include/boost/url/impl/url_base.hpp

99.5% Lines (1514/1522) 100.0% Functions (80/80)
include/boost/url/impl/url_base.hpp
Line TLA Hits Source Code
1 //
2 // Copyright (c) 2019 Vinnie Falco (vinnie.falco@gmail.com)
3 // Copyright (c) 2022 Alan de Freitas (alandefreitas@gmail.com)
4 //
5 // Distributed under the Boost Software License, Version 1.0. (See accompanying
6 // file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
7 //
8 // Official repository: https://github.com/boostorg/url
9 //
10
11 #ifndef BOOST_URL_IMPL_URL_BASE_HPP
12 #define BOOST_URL_IMPL_URL_BASE_HPP
13
14 #include <boost/url/encode.hpp>
15 #include <boost/url/error.hpp>
16 #include <boost/url/host_type.hpp>
17 #include <boost/url/scheme.hpp>
18 #include <boost/url/url_view.hpp>
19 #include <boost/url/detail/any_params_iter.hpp>
20 #include <boost/url/detail/any_segments_iter.hpp>
21 #include <boost/url/detail/decode.hpp>
22 #include <boost/url/detail/encode.hpp>
23 #include <boost/url/detail/except.hpp>
24 #include <boost/url/detail/normalize.hpp>
25 #include <boost/url/detail/path.hpp>
26 #include <boost/url/detail/print.hpp>
27 #include <boost/url/grammar/ci_string.hpp>
28 #include <boost/url/rfc/authority_rule.hpp>
29 #include <boost/url/rfc/query_rule.hpp>
30 #include <boost/url/rfc/ipv6_address_rule.hpp>
31 #include <boost/url/rfc/detail/charsets.hpp>
32 #include <boost/url/rfc/detail/host_rule.hpp>
33 #include <boost/url/rfc/detail/ipvfuture_rule.hpp>
34 #include <boost/url/rfc/detail/path_rules.hpp>
35 #include <boost/url/rfc/detail/port_rule.hpp>
36 #include <boost/url/rfc/detail/scheme_rule.hpp>
37 #include <boost/url/rfc/detail/userinfo_rule.hpp>
38 #include <boost/url/grammar/parse.hpp>
39 #include <boost/url/detail/move_chars.hpp>
40 #include <cstring>
41 #include <iostream>
42 #include <stdexcept>
43 #include <utility>
44
45 namespace boost {
46 namespace urls {
47
48 //------------------------------------------------
49
50 // these objects help handle the cases
51 // where the user passes in strings that
52 // come from inside the url buffer.
53
54 inline
55 18176 url_base::
56 op_t::
57 ~op_t()
58 {
59 18176 if(old)
60 2561 u.cleanup(*this);
61 18176 u.check_invariants();
62 18176 }
63
64 inline
65 18176 url_base::
66 op_t::
67 op_t(
68 url_base& impl_,
69 core::string_view* s0_,
70 18176 core::string_view* s1_) noexcept
71 18176 : u(impl_)
72 18176 , s0(s0_)
73 18176 , s1(s1_)
74 {
75 18176 u.check_invariants();
76 18176 }
77
78 inline
79 void
80 9335 url_base::
81 op_t::
82 move(
83 char* dest,
84 char const* src,
85 std::size_t n) noexcept
86 {
87 9335 if(! n)
88 2519 return;
89 6816 if(s0)
90 {
91 4885 if(s1)
92 654 return detail::move_chars(
93 654 dest, src, n, *s0, *s1);
94 4231 return detail::move_chars(
95 4231 dest, src, n, *s0);
96 }
97 1931 detail::move_chars(
98 dest, src, n);
99 }
100
101 //------------------------------------------------
102
103 // construct reference
104 inline
105 1647 url_base::
106 url_base(
107 1647 detail::url_impl const& impl) noexcept
108 1647 : url_view_base(impl)
109 {
110 1647 }
111
112 inline
113 void
114 305 url_base::
115 reserve_impl(std::size_t n)
116 {
117 305 op_t op(*this);
118 305 reserve_impl(n, op);
119 304 if(s_)
120 302 s_[size()] = '\0';
121 305 }
122
123 // make a copy of u
124 inline
125 void
126 5189 url_base::
127 copy(url_view_base const& u)
128 {
129 5189 if (this == &u)
130 140 return;
131 5175 op_t op(*this);
132 5175 if(u.size() == 0)
133 {
134 126 clear();
135 126 return;
136 }
137 5049 reserve_impl(
138 u.size(), op);
139 5046 impl_ = u.impl();
140 5046 impl_.cs_ = s_;
141 5046 impl_.from_ = {from::url};
142 5046 std::memcpy(s_,
143 5046 u.data(), u.size());
144 5046 s_[size()] = '\0';
145 5175 }
146
147 //------------------------------------------------
148 //
149 // Scheme
150 //
151 //------------------------------------------------
152
153 inline
154 url_base&
155 376 url_base::
156 set_scheme(core::string_view s)
157 {
158 376 set_scheme_impl(
159 s, string_to_scheme(s));
160 359 return *this;
161 }
162
163 inline
164 url_base&
165 182 url_base::
166 set_scheme_id(urls::scheme id)
167 {
168 182 if(id == urls::scheme::unknown)
169 14 detail::throw_invalid_argument();
170 168 if(id == urls::scheme::none)
171 18 return remove_scheme();
172 150 set_scheme_impl(to_string(id), id);
173 119 return *this;
174 }
175
176 inline
177 url_base&
178 53 url_base::
179 remove_scheme()
180 {
181 53 op_t op(*this);
182 53 auto const sn = impl_.len(id_scheme);
183 53 if(sn == 0)
184 15 return *this;
185 38 auto const po = impl_.offset(id_path);
186 38 auto fseg = first_segment();
187 bool const encode_colon =
188 38 !has_authority() &&
189 21 impl_.nseg_ > 0 &&
190 70 s_[po] != '/' &&
191 11 fseg.contains(':');
192 38 if(!encode_colon)
193 {
194 // just remove the scheme
195 29 resize_impl(id_scheme, 0, op);
196 29 impl_.scheme_ = urls::scheme::none;
197 29 check_invariants();
198 29 return *this;
199 }
200 // encode any ":" in the first path segment
201 9 BOOST_ASSERT(sn >= 2);
202 9 auto pn = impl_.len(id_path);
203 9 std::size_t cn = 0;
204 46 for (char c: fseg)
205 37 cn += c == ':';
206 std::size_t new_size =
207 9 size() - sn + 2 * cn;
208 9 bool need_resize = new_size > size();
209 9 if (need_resize)
210 {
211 1 resize_impl(
212 1 id_path, pn + 2 * cn, op);
213 }
214 // move [id_scheme, id_path) left
215 9 op.move(
216 s_,
217 9 s_ + sn,
218 po - sn);
219 // move [id_path, id_query) left
220 9 auto qo = impl_.offset(id_query);
221 9 op.move(
222 9 s_ + po - sn,
223 9 s_ + po,
224 qo - po);
225 // move [id_query, id_end) left
226 9 op.move(
227 9 s_ + qo - sn + 2 * cn,
228 9 s_ + qo,
229 9 impl_.offset(id_end) - qo);
230
231 // adjust part offsets.
232 // (po and qo are invalidated)
233 9 if (need_resize)
234 {
235 1 impl_.adjust_left(id_user, id_end, sn);
236 }
237 else
238 {
239 8 impl_.adjust_left(id_user, id_path, sn);
240 8 impl_.adjust_left(id_query, id_end, sn - 2 * cn);
241 }
242 9 if (encode_colon)
243 {
244 // move the 2nd, 3rd, ... segments
245 9 auto begin = s_ + impl_.offset(id_path);
246 9 auto it = begin;
247 9 auto end = begin + pn;
248 46 while (it != end &&
249 40 *it != '/')
250 37 ++it;
251 // we don't need op here because this is
252 // an internal operation
253 9 std::memmove(it + (2 * cn), it, end - it);
254
255 // move 1st segment
256 9 auto src = s_ + impl_.offset(id_path) + pn;
257 9 auto dest = s_ + impl_.offset(id_query);
258 9 src -= end - it;
259 9 dest -= end - it;
260 9 pn -= end - it;
261 do {
262 37 --src;
263 37 --dest;
264 37 if (*src != ':')
265 {
266 25 *dest = *src;
267 }
268 else
269 {
270 // use uppercase as required by
271 // syntax-based normalization
272 12 *dest-- = 'A';
273 12 *dest-- = '3';
274 12 *dest = '%';
275 }
276 37 --pn;
277 37 } while (pn);
278 }
279 9 s_[size()] = '\0';
280 9 impl_.scheme_ = urls::scheme::none;
281 9 return *this;
282 53 }
283
284 //------------------------------------------------
285 //
286 // Authority
287 //
288 //------------------------------------------------
289
290 inline
291 url_base&
292 120 url_base::
293 set_encoded_authority(
294 pct_string_view s)
295 {
296 120 op_t op(*this, &detail::ref(s));
297 122 authority_view a = grammar::parse(
298 s, authority_rule
299 120 ).value(BOOST_URL_POS);
300 119 auto n = s.size() + 2;
301 auto const need_slash =
302 141 ! is_path_absolute() &&
303 22 impl_.len(id_path) > 0;
304 119 if(need_slash)
305 2 ++n;
306 119 auto dest = resize_impl(
307 id_user, id_path, n, op);
308 119 dest[0] = '/';
309 119 dest[1] = '/';
310 119 std::memcpy(dest + 2,
311 119 s.data(), s.size());
312 119 if(need_slash)
313 2 dest[n - 1] = '/';
314 119 impl_.apply_authority(a.u_);
315 119 if(need_slash)
316 2 impl_.adjust_right(
317 id_query, id_end, 1);
318 119 return *this;
319 120 }
320
321 inline
322 url_base&
323 259 url_base::
324 remove_authority()
325 {
326 259 if(! has_authority())
327 70 return *this;
328
329 189 op_t op(*this);
330 189 auto path = impl_.get(id_path);
331 189 bool const need_dot = path.starts_with("//");
332 189 if(need_dot)
333 {
334 // prepend "/.", can't throw
335 5 auto p = resize_impl(
336 id_user, id_path, 2, op);
337 5 p[0] = '/';
338 5 p[1] = '.';
339 5 impl_.split(id_user, 0);
340 5 impl_.split(id_pass, 0);
341 5 impl_.split(id_host, 0);
342 5 impl_.split(id_port, 0);
343 }
344 else
345 {
346 184 resize_impl(
347 id_user, id_path, 0, op);
348 }
349 189 impl_.host_type_ =
350 urls::host_type::none;
351 189 return *this;
352 189 }
353
354 //------------------------------------------------
355 //
356 // Userinfo
357 //
358 //------------------------------------------------
359
360 inline
361 url_base&
362 225 url_base::
363 set_userinfo(
364 core::string_view s)
365 {
366 225 op_t op(*this, &s);
367 225 encoding_opts opt;
368 225 auto const n = encoded_size(
369 s, detail::userinfo_chars, opt);
370 225 auto dest = set_userinfo_impl(n, op);
371 225 encode(
372 dest,
373 n,
374 s,
375 detail::userinfo_chars,
376 opt);
377 225 auto const pos = impl_.get(
378 id_user, id_host
379 225 ).find_first_of(':');
380 225 if(pos != core::string_view::npos)
381 {
382 22 impl_.split(id_user, pos);
383 // find ':' in plain string
384 auto const pos2 =
385 22 s.find_first_of(':');
386 22 if(pos2 != core::string_view::npos)
387 {
388 // pos2 is the ':' index in plain input (user[:pass])
389 // decoded user is [0, pos2), decoded pass is (pos2, end].
390 22 impl_.decoded_[id_user] =
391 22 detail::to_size_type(pos2);
392 22 impl_.decoded_[id_pass] =
393 22 detail::to_size_type(s.size() - pos2 - 1);
394 }
395 else
396 {
397 impl_.decoded_[id_user] =
398 detail::to_size_type(s.size());
399 impl_.decoded_[id_pass] = 0;
400 }
401 }
402 else
403 {
404 203 impl_.decoded_[id_user] =
405 203 detail::to_size_type(s.size());
406 203 impl_.decoded_[id_pass] = 0;
407 }
408 225 return *this;
409 225 }
410
411 inline
412 url_base&
413 52 url_base::
414 set_encoded_userinfo(
415 pct_string_view s)
416 {
417 52 op_t op(*this, &detail::ref(s));
418 52 auto const pos = s.find_first_of(':');
419 52 if(pos != core::string_view::npos)
420 {
421 // user:pass
422 7 auto const s0 = s.substr(0, pos);
423 7 auto const s1 = s.substr(pos + 1);
424 auto const n0 =
425 7 detail::re_encoded_size_unsafe(
426 s0,
427 detail::user_chars);
428 auto const n1 =
429 7 detail::re_encoded_size_unsafe(s1,
430 detail::password_chars);
431 auto dest =
432 7 set_userinfo_impl(n0 + n1 + 1, op);
433 7 impl_.decoded_[id_user] =
434 7 detail::to_size_type(detail::re_encode_unsafe(
435 dest,
436 7 dest + n0,
437 s0,
438 detail::user_chars));
439 7 *dest++ = ':';
440 7 impl_.decoded_[id_pass] =
441 7 detail::to_size_type(detail::re_encode_unsafe(
442 dest,
443 7 dest + n1,
444 s1,
445 detail::password_chars));
446 7 impl_.split(id_user, 2 + n0);
447 }
448 else
449 {
450 // user
451 auto const n =
452 45 detail::re_encoded_size_unsafe(
453 s, detail::user_chars);
454 45 auto dest = set_userinfo_impl(n, op);
455 45 impl_.decoded_[id_user] =
456 90 detail::to_size_type(detail::re_encode_unsafe(
457 dest,
458 45 dest + n,
459 s,
460 detail::user_chars));
461 45 impl_.split(id_user, 2 + n);
462 45 impl_.decoded_[id_pass] = 0;
463 }
464 52 return *this;
465 52 }
466
467 inline
468 url_base&
469 214 url_base::
470 remove_userinfo() noexcept
471 {
472 214 if(impl_.len(id_pass) == 0)
473 138 return *this; // no userinfo
474
475 76 op_t op(*this);
476 // keep authority '//'
477 76 resize_impl(
478 id_user, id_host, 2, op);
479 76 impl_.decoded_[id_user] = 0;
480 76 impl_.decoded_[id_pass] = 0;
481 76 return *this;
482 76 }
483
484 //------------------------------------------------
485
486 inline
487 url_base&
488 241 url_base::
489 set_user(core::string_view s)
490 {
491 241 op_t op(*this, &s);
492 241 encoding_opts opt;
493 241 auto const n = encoded_size(
494 s, detail::user_chars, opt);
495 241 auto dest = set_user_impl(n, op);
496 241 encode_unsafe(
497 dest,
498 n,
499 s,
500 detail::user_chars,
501 opt);
502 241 impl_.decoded_[id_user] =
503 241 detail::to_size_type(s.size());
504 241 return *this;
505 241 }
506
507 inline
508 url_base&
509 147 url_base::
510 set_encoded_user(
511 pct_string_view s)
512 {
513 147 op_t op(*this, &detail::ref(s));
514 auto const n =
515 147 detail::re_encoded_size_unsafe(
516 s, detail::user_chars);
517 147 auto dest = set_user_impl(n, op);
518 147 impl_.decoded_[id_user] =
519 294 detail::to_size_type(detail::re_encode_unsafe(
520 dest,
521 147 dest + n,
522 s,
523 detail::user_chars));
524 147 BOOST_ASSERT(
525 impl_.decoded_[id_user] ==
526 s.decoded_size());
527 147 return *this;
528 147 }
529
530 //------------------------------------------------
531
532 inline
533 url_base&
534 231 url_base::
535 set_password(core::string_view s)
536 {
537 231 op_t op(*this, &s);
538 231 encoding_opts opt;
539 231 auto const n = encoded_size(
540 s, detail::password_chars, opt);
541 231 auto dest = set_password_impl(n, op);
542 231 encode_unsafe(
543 dest,
544 n,
545 s,
546 detail::password_chars,
547 opt);
548 231 impl_.decoded_[id_pass] =
549 231 detail::to_size_type(s.size());
550 231 return *this;
551 231 }
552
553 inline
554 url_base&
555 140 url_base::
556 set_encoded_password(
557 pct_string_view s)
558 {
559 140 op_t op(*this, &detail::ref(s));
560 auto const n =
561 140 detail::re_encoded_size_unsafe(
562 s,
563 detail::password_chars);
564 140 auto dest = set_password_impl(n, op);
565 140 impl_.decoded_[id_pass] =
566 280 detail::to_size_type(detail::re_encode_unsafe(
567 dest,
568 140 dest + n,
569 s,
570 detail::password_chars));
571 140 BOOST_ASSERT(
572 impl_.decoded_[id_pass] ==
573 s.decoded_size());
574 140 return *this;
575 140 }
576
577 inline
578 url_base&
579 19 url_base::
580 remove_password() noexcept
581 {
582 19 auto const n = impl_.len(id_pass);
583 19 if(n < 2)
584 12 return *this; // no password
585
586 7 op_t op(*this);
587 // clear password, retain '@'
588 auto dest =
589 7 resize_impl(id_pass, 1, op);
590 7 dest[0] = '@';
591 7 impl_.decoded_[id_pass] = 0;
592 7 return *this;
593 7 }
594
595 //------------------------------------------------
596 //
597 // Host
598 //
599 //------------------------------------------------
600 /*
601 host_type host_type() // ipv4, ipv6, ipvfuture, name
602
603 std::string host() // return encoded_host().decode()
604 pct_string_view encoded_host() // return host part, as-is
605 std::string host_address() // return encoded_host_address().decode()
606 pct_string_view encoded_host_address() // ipv4, ipv6, ipvfut, or encoded name, no brackets
607
608 ipv4_address host_ipv4_address() // return ipv4_address or {}
609 ipv6_address host_ipv6_address() // return ipv6_address or {}
610 core::string_view host_ipvfuture() // return ipvfuture or {}
611 std::string host_name() // return decoded name or ""
612 pct_string_view encoded_host_name() // return encoded host name or ""
613
614 --------------------------------------------------
615
616 set_host( core::string_view ) // set host part from plain text
617 set_encoded_host( pct_string_view ) // set host part from encoded text
618 set_host_address( core::string_view ) // set host from ipv4, ipv6, ipvfut, or plain reg-name string
619 set_encoded_host_address( pct_string_view ) // set host from ipv4, ipv6, ipvfut, or encoded reg-name string
620
621 set_host_ipv4( ipv4_address ) // set ipv4
622 set_host_ipv6( ipv6_address ) // set ipv6
623 set_host_ipvfuture( core::string_view ) // set ipvfuture
624 set_host_name( core::string_view ) // set name from plain
625 set_encoded_host_name( pct_string_view ) // set name from encoded
626 */
627
628 // set host part from plain text
629 inline
630 url_base&
631 401 url_base::
632 set_host(
633 core::string_view s)
634 {
635 401 if( s.size() > 2 &&
636 450 s.front() == '[' &&
637 49 s.back() == ']')
638 {
639 // IP-literal
640 49 if (s[1] != 'v')
641 {
642 // IPv6-address
643 48 auto innersv = s.substr(1, s.size() - 2);
644 48 auto innerit = innersv.begin();
645 48 auto endit = innersv.end();
646 48 auto rv = grammar::parse(
647 innerit,
648 endit,
649 ipv6_address_rule);
650 48 if(rv)
651 {
652 48 if (innerit == endit)
653 {
654 47 set_host_ipv6_and_encoded_zone_id(*rv, {});
655 48 return *this;
656 }
657 // IPv6addrz: https://datatracker.ietf.org/doc/html/rfc6874
658 1 auto chars_left = endit - innerit;
659 2 if (chars_left >= 2 &&
660 1 *innerit++ == '%')
661 {
662 1 core::string_view zone_id_str = {&*innerit, std::size_t(chars_left - 1)};
663 1 set_host_ipv6_and_zone_id(*rv, zone_id_str);
664 1 return *this;
665 }
666 }
667 }
668 else
669 {
670 // IPvFuture
671 1 auto rv = grammar::parse(
672 1 s.substr(1, s.size() - 2),
673 detail::ipvfuture_rule);
674 1 if(rv)
675 1 return set_host_ipvfuture(rv->str);
676 }
677 }
678 352 else if(s.size() >= 7) // "0.0.0.0"
679 {
680 // IPv4-address
681 348 auto rv = parse_ipv4_address(s);
682 348 if(rv)
683 189 return set_host_ipv4(*rv);
684 }
685
686 // reg-name
687 163 op_t op(*this, &s);
688 163 encoding_opts opt;
689 163 auto const n = encoded_size(
690 s, detail::host_chars, opt);
691 163 auto dest = set_host_impl(n, op);
692 163 encode(
693 dest,
694 163 impl_.get(id_path).data() - dest,
695 s,
696 detail::host_chars,
697 opt);
698 163 impl_.decoded_[id_host] =
699 163 detail::to_size_type(s.size());
700 163 impl_.host_type_ =
701 urls::host_type::name;
702 163 return *this;
703 163 }
704
705 // set host part from encoded text
706 inline
707 url_base&
708 218 url_base::
709 set_encoded_host(
710 pct_string_view s)
711 {
712 218 if( s.size() > 2 &&
713 235 s.front() == '[' &&
714 17 s.back() == ']')
715 {
716 // IP-literal
717 17 if (s[1] != 'v')
718 {
719 // IPv6-address
720 16 auto innersv = s.substr(1, s.size() - 2);
721 16 auto innerit = innersv.begin();
722 16 auto endit = innersv.end();
723 16 auto rv = grammar::parse(
724 innerit,
725 endit,
726 ipv6_address_rule);
727 16 if(rv)
728 {
729 8 if (innerit == endit)
730 {
731 5 set_host_ipv6_and_encoded_zone_id(*rv, {});
732 6 return *this;
733 }
734 // IPv6addrz: https://datatracker.ietf.org/doc/html/rfc6874
735 3 auto chars_left = endit - innerit;
736 4 if (chars_left >= 3 &&
737 1 *innerit++ == '%' &&
738 5 *innerit++ == '2' &&
739 1 *innerit++ == '5')
740 {
741 1 auto const nz = std::size_t(chars_left - 3);
742 1 core::string_view zone_id_str = {&*innerit, std::size_t(chars_left - 3)};
743 1 std::size_t dnz = detail::decode_bytes_unsafe(zone_id_str);
744 1 pct_string_view zone_id_pct = make_pct_string_view_unsafe(innerit, nz, dnz);
745 1 set_host_ipv6_and_encoded_zone_id(*rv, zone_id_pct);
746 1 return *this;
747 }
748 }
749 }
750 else
751 {
752 // IPvFuture
753 1 auto rv = grammar::parse(
754 1 s.substr(1, s.size() - 2),
755 detail::ipvfuture_rule);
756 1 if(rv)
757 1 return set_host_ipvfuture(rv->str);
758 }
759 }
760 201 else if(s.size() >= 7) // "0.0.0.0"
761 {
762 // IPv4-address
763 74 auto rv = parse_ipv4_address(s);
764 74 if(rv)
765 5 return set_host_ipv4(*rv);
766 }
767
768 // reg-name
769 206 op_t op(*this, &detail::ref(s));
770 206 auto const n = detail::re_encoded_size_unsafe(
771 s, detail::host_chars);
772 206 auto dest = set_host_impl(n, op);
773 206 impl_.decoded_[id_host] =
774 412 detail::to_size_type(detail::re_encode_unsafe(
775 dest,
776 206 impl_.get(id_path).data(),
777 s,
778 detail::host_chars));
779 206 BOOST_ASSERT(impl_.decoded_[id_host] ==
780 s.decoded_size());
781 206 impl_.host_type_ =
782 urls::host_type::name;
783 206 return *this;
784 206 }
785
786 inline
787 url_base&
788 10 url_base::
789 set_host_address(
790 core::string_view s)
791 {
792 10 if (!s.empty())
793 {
794 // IP-literal
795 9 if (s[0] != 'v')
796 {
797 // IPv6-address
798 8 auto innerit = s.begin();
799 8 auto endit = s.end();
800 8 auto rv = grammar::parse(
801 innerit,
802 endit,
803 ipv6_address_rule);
804 8 if(rv)
805 {
806 2 if (innerit == endit)
807 {
808 1 set_host_ipv6_and_encoded_zone_id(*rv, {});
809 2 return *this;
810 }
811 // IPv6addrz: https://datatracker.ietf.org/doc/html/rfc6874
812 1 auto chars_left = endit - innerit;
813 2 if (chars_left >= 2 &&
814 1 *innerit++ == '%')
815 {
816 1 core::string_view zone_id_str = {&*innerit, std::size_t(chars_left - 1)};
817 1 set_host_ipv6_and_zone_id(*rv, zone_id_str);
818 1 return *this;
819 }
820 }
821 }
822
823 // IPvFuture
824 7 auto rv = grammar::parse(s, detail::ipvfuture_rule);
825 7 if(rv)
826 1 return set_host_ipvfuture(rv->str);
827
828 6 if(s.size() >= 7) // "0.0.0.0"
829 {
830 // IPv4-address
831 5 auto rv2 = parse_ipv4_address(s);
832 5 if(rv2)
833 2 return set_host_ipv4(*rv2);
834 }
835 }
836
837 // reg-name
838 5 op_t op(*this, &s);
839 5 encoding_opts opt;
840 5 auto const n = encoded_size(
841 s, detail::host_chars, opt);
842 5 auto dest = set_host_impl(n, op);
843 5 encode(
844 dest,
845 5 impl_.get(id_path).data() - dest,
846 s,
847 detail::host_chars,
848 opt);
849 5 impl_.decoded_[id_host] =
850 5 detail::to_size_type(s.size());
851 5 impl_.host_type_ =
852 urls::host_type::name;
853 5 return *this;
854 5 }
855
856 inline
857 url_base&
858 8 url_base::
859 set_encoded_host_address(
860 pct_string_view s)
861 {
862 8 if( !s.empty() )
863 {
864 // IP-literal
865 7 if (s[0] != 'v')
866 {
867 // IPv6-address
868 6 auto innerit = s.begin();
869 6 auto endit = s.end();
870 6 auto rv = grammar::parse(
871 innerit,
872 endit,
873 ipv6_address_rule);
874 6 if(rv)
875 {
876 2 if (innerit == endit)
877 {
878 1 set_host_ipv6_and_encoded_zone_id(*rv, {});
879 3 return *this;
880 }
881 // IPv6addrz: https://datatracker.ietf.org/doc/html/rfc6874
882 1 auto chars_left = endit - innerit;
883 2 if (chars_left >= 3 &&
884 1 *innerit++ == '%' &&
885 3 *innerit++ == '2' &&
886 1 *innerit++ == '5')
887 {
888 1 auto const nz = std::size_t(chars_left - 3);
889 1 core::string_view zone_id_str = {&*innerit, std::size_t(chars_left - 3)};
890 1 std::size_t dnz = detail::decode_bytes_unsafe(zone_id_str);
891 1 pct_string_view zone_id_pct = make_pct_string_view_unsafe(innerit, nz, dnz);
892 1 set_host_ipv6_and_encoded_zone_id(*rv, zone_id_pct);
893 1 return *this;
894 }
895 }
896
897 4 if(s.size() >= 7) // "0.0.0.0"
898 {
899 // IPv4-address
900 3 auto rv2 = parse_ipv4_address(s);
901 3 if(rv2)
902 1 return set_host_ipv4(*rv2);
903 }
904 }
905 else
906 {
907 // IPvFuture
908 1 auto rv = grammar::parse(
909 s, detail::ipvfuture_rule);
910 1 if(rv)
911 1 return set_host_ipvfuture(rv->str);
912 }
913 }
914
915 // reg-name
916 4 op_t op(*this, &detail::ref(s));
917 4 auto const n = detail::re_encoded_size_unsafe(
918 s, detail::host_chars);
919 4 auto dest = set_host_impl(n, op);
920 4 impl_.decoded_[id_host] =
921 8 detail::to_size_type(detail::re_encode_unsafe(
922 dest,
923 4 impl_.get(id_path).data(),
924 s,
925 detail::host_chars));
926 4 BOOST_ASSERT(impl_.decoded_[id_host] ==
927 s.decoded_size());
928 4 impl_.host_type_ =
929 urls::host_type::name;
930 4 return *this;
931 4 }
932
933 inline
934 url_base&
935 203 url_base::
936 set_host_ipv4(
937 ipv4_address const& addr)
938 {
939 203 op_t op(*this);
940 char buf[urls::ipv4_address::max_str_len];
941 203 auto s = addr.to_buffer(buf, sizeof(buf));
942 203 auto dest = set_host_impl(s.size(), op);
943 203 std::memcpy(dest, s.data(), s.size());
944 203 impl_.decoded_[id_host] =
945 203 detail::to_size_type(impl_.len(id_host));
946 203 impl_.host_type_ = urls::host_type::ipv4;
947 203 auto bytes = addr.to_bytes();
948 203 std::memcpy(
949 203 impl_.ip_addr_,
950 203 bytes.data(),
951 bytes.size());
952 203 return *this;
953 203 }
954
955 inline
956 url_base&
957 5 url_base::
958 set_host_ipv6(
959 ipv6_address const& addr)
960 {
961 5 set_host_ipv6_and_encoded_zone_id(addr, encoded_zone_id());
962 5 return *this;
963 }
964
965 inline
966 url_base&
967 3 url_base::
968 set_zone_id(core::string_view s)
969 {
970 3 set_host_ipv6_and_zone_id(host_ipv6_address(), s);
971 3 return *this;
972 }
973
974 inline
975 url_base&
976 3 url_base::
977 set_encoded_zone_id(pct_string_view s)
978 {
979 3 set_host_ipv6_and_encoded_zone_id(host_ipv6_address(), s);
980 3 return *this;
981 }
982
983 inline
984 void
985 5 url_base::
986 set_host_ipv6_and_zone_id(
987 ipv6_address const& addr,
988 core::string_view zone_id)
989 {
990 5 op_t op(*this, &zone_id);
991 char ipv6_str_buf[urls::ipv6_address::max_str_len];
992 5 auto ipv6_str = addr.to_buffer(ipv6_str_buf, sizeof(ipv6_str_buf));
993 5 bool const has_zone_id = !zone_id.empty();
994 5 encoding_opts opt;
995 5 auto const ipn = ipv6_str.size();
996 5 auto const zn = encoded_size(zone_id, unreserved_chars, opt);
997 5 auto const n = ipn + 2 + has_zone_id * (3 + zn);
998 5 auto dest = set_host_impl(n, op);
999 5 *dest++ = '[';
1000 5 std::memcpy(dest, ipv6_str.data(), ipn);
1001 5 dest += ipn;
1002 5 if (has_zone_id)
1003 {
1004 5 *dest++ = '%';
1005 5 *dest++ = '2';
1006 5 *dest++ = '5';
1007 5 encode(dest, zn, zone_id, unreserved_chars, opt);
1008 5 dest += zn;
1009 }
1010 5 *dest++ = ']';
1011 // ipn + |"["| + |"]"| + (has_zone_id ? |"%"| + zn : 0)
1012 10 impl_.decoded_[id_host] = detail::to_size_type(
1013 5 ipn + 2 + has_zone_id * (1 + zone_id.size()));
1014 5 impl_.host_type_ = urls::host_type::ipv6;
1015 5 auto bytes = addr.to_bytes();
1016 5 std::memcpy(
1017 5 impl_.ip_addr_,
1018 5 bytes.data(),
1019 bytes.size());
1020 5 }
1021
1022 inline
1023 void
1024 64 url_base::
1025 set_host_ipv6_and_encoded_zone_id(
1026 ipv6_address const& addr,
1027 pct_string_view zone_id)
1028 {
1029 64 op_t op(*this, &detail::ref(zone_id));
1030 char ipv6_str_buf[urls::ipv6_address::max_str_len];
1031 64 auto ipv6_str = addr.to_buffer(ipv6_str_buf, sizeof(ipv6_str_buf));
1032 64 bool const has_zone_id = !zone_id.empty();
1033 64 auto const ipn = ipv6_str.size();
1034 64 auto const zn = detail::re_encoded_size_unsafe(zone_id, unreserved_chars);
1035 64 auto const n = ipn + 2 + has_zone_id * (3 + zn);
1036 64 auto dest = set_host_impl(n, op);
1037 64 *dest++ = '[';
1038 64 std::memcpy(dest, ipv6_str.data(), ipn);
1039 64 dest += ipn;
1040 64 std::size_t dzn = 0;
1041 64 if (has_zone_id)
1042 {
1043 6 *dest++ = '%';
1044 6 *dest++ = '2';
1045 6 *dest++ = '5';
1046 6 dzn = detail::re_encode_unsafe(dest, dest + zn, zone_id, unreserved_chars);
1047 }
1048 64 *dest++ = ']';
1049 // ipn + |"["| + |"]"| + (has_zone_id ? |"%"| + zn : 0)
1050 128 impl_.decoded_[id_host] = detail::to_size_type(
1051 64 ipn + 2 + has_zone_id * (1 + dzn));
1052 64 impl_.host_type_ = urls::host_type::ipv6;
1053 64 auto bytes = addr.to_bytes();
1054 64 std::memcpy(
1055 64 impl_.ip_addr_,
1056 64 bytes.data(),
1057 bytes.size());
1058 64 }
1059
1060 inline
1061 url_base&
1062 7 url_base::
1063 set_host_ipvfuture(
1064 core::string_view s)
1065 {
1066 7 op_t op(*this, &s);
1067 // validate
1068 8 grammar::parse(s,
1069 detail::ipvfuture_rule
1070 8 ).value(BOOST_URL_POS);
1071 6 auto dest = set_host_impl(
1072 6 s.size() + 2, op);
1073 6 *dest++ = '[';
1074 6 dest += s.copy(dest, s.size());
1075 6 *dest = ']';
1076 6 impl_.host_type_ =
1077 urls::host_type::ipvfuture;
1078 6 impl_.decoded_[id_host] =
1079 6 detail::to_size_type(s.size() + 2);
1080 6 return *this;
1081 7 }
1082
1083 inline
1084 url_base&
1085 4 url_base::
1086 set_host_name(
1087 core::string_view s)
1088 {
1089 4 bool is_ipv4 = false;
1090 4 if(s.size() >= 7) // "0.0.0.0"
1091 {
1092 // IPv4-address
1093 3 if(parse_ipv4_address(s).has_value())
1094 1 is_ipv4 = true;
1095 }
1096 4 auto allowed = detail::host_chars;
1097 4 if(is_ipv4)
1098 1 allowed = allowed - '.';
1099
1100 4 op_t op(*this, &s);
1101 4 encoding_opts opt;
1102 4 auto const n = encoded_size(
1103 s, allowed, opt);
1104 4 auto dest = set_host_impl(n, op);
1105 4 encode_unsafe(
1106 dest,
1107 n,
1108 s,
1109 allowed,
1110 opt);
1111 4 impl_.host_type_ =
1112 urls::host_type::name;
1113 4 impl_.decoded_[id_host] =
1114 4 detail::to_size_type(s.size());
1115 4 return *this;
1116 4 }
1117
1118 inline
1119 url_base&
1120 4 url_base::
1121 set_encoded_host_name(
1122 pct_string_view s)
1123 {
1124 4 bool is_ipv4 = false;
1125 4 if(s.size() >= 7) // "0.0.0.0"
1126 {
1127 // IPv4-address
1128 3 if(parse_ipv4_address(s).has_value())
1129 1 is_ipv4 = true;
1130 }
1131 4 auto allowed = detail::host_chars;
1132 4 if(is_ipv4)
1133 1 allowed = allowed - '.';
1134
1135 4 op_t op(*this, &detail::ref(s));
1136 4 auto const n = detail::re_encoded_size_unsafe(
1137 s, allowed);
1138 4 auto dest = set_host_impl(n, op);
1139 4 impl_.decoded_[id_host] =
1140 8 detail::to_size_type(detail::re_encode_unsafe(
1141 dest,
1142 4 dest + n,
1143 s,
1144 allowed));
1145 4 BOOST_ASSERT(
1146 impl_.decoded_[id_host] ==
1147 s.decoded_size());
1148 4 impl_.host_type_ =
1149 urls::host_type::name;
1150 4 return *this;
1151 4 }
1152
1153 //------------------------------------------------
1154
1155 inline
1156 url_base&
1157 215 url_base::
1158 set_port_number(
1159 std::uint16_t n)
1160 {
1161 215 op_t op(*this);
1162 auto s =
1163 215 detail::make_printed(n);
1164 215 auto dest = set_port_impl(
1165 215 s.string().size(), op);
1166 215 std::memcpy(
1167 215 dest, s.string().data(),
1168 215 s.string().size());
1169 215 impl_.port_number_ = n;
1170 215 return *this;
1171 215 }
1172
1173 inline
1174 url_base&
1175 389 url_base::
1176 set_port(
1177 core::string_view s)
1178 {
1179 389 op_t op(*this, &s);
1180 410 auto t = grammar::parse(s,
1181 21 detail::port_rule{}
1182 389 ).value(BOOST_URL_POS);
1183 auto dest =
1184 368 set_port_impl(t.str.size(), op);
1185 368 std::memcpy(dest,
1186 368 t.str.data(), t.str.size());
1187 368 if(t.has_number)
1188 208 impl_.port_number_ = t.number;
1189 else
1190 160 impl_.port_number_ = 0;
1191 368 return *this;
1192 389 }
1193
1194 inline
1195 url_base&
1196 206 url_base::
1197 remove_port() noexcept
1198 {
1199 206 op_t op(*this);
1200 206 resize_impl(id_port, 0, op);
1201 206 impl_.port_number_ = 0;
1202 412 return *this;
1203 206 }
1204
1205 //------------------------------------------------
1206 //
1207 // Compound Fields
1208 //
1209 //------------------------------------------------
1210
1211 inline
1212 url_base&
1213 14 url_base::
1214 remove_origin()
1215 {
1216 // these two calls perform 2 memmoves instead of 1
1217 14 remove_authority();
1218 14 remove_scheme();
1219 14 return *this;
1220 }
1221
1222 //------------------------------------------------
1223 //
1224 // Path
1225 //
1226 //------------------------------------------------
1227
1228 inline
1229 bool
1230 52 url_base::
1231 set_path_absolute(
1232 bool absolute)
1233 {
1234 52 op_t op(*this);
1235
1236 // check if path empty
1237 52 if(impl_.len(id_path) == 0)
1238 {
1239 40 if(! absolute)
1240 {
1241 // already not absolute
1242 34 return true;
1243 }
1244
1245 // add '/'
1246 6 auto dest = resize_impl(
1247 id_path, 1, op);
1248 6 *dest = '/';
1249 6 ++impl_.decoded_[id_path];
1250 6 return true;
1251 }
1252
1253 // check if path absolute
1254 12 if(s_[impl_.offset(id_path)] == '/')
1255 {
1256 9 if(absolute)
1257 {
1258 // already absolute
1259 2 return true;
1260 }
1261
1262 11 if( has_authority() &&
1263 4 impl_.len(id_path) > 1)
1264 {
1265 // can't do it, paths are always
1266 // absolute when authority present!
1267 2 return false;
1268 }
1269
1270 5 auto p = encoded_path();
1271 5 auto pos = p.find_first_of(":/", 1);
1272 6 if (pos != core::string_view::npos &&
1273 1 p[pos] == ':')
1274 {
1275 // prepend with .
1276 1 auto n = impl_.len(id_path);
1277 1 resize_impl(id_path, n + 1, op);
1278 1 std::memmove(
1279 2 s_ + impl_.offset(id_path) + 1,
1280 1 s_ + impl_.offset(id_path), n);
1281 1 *(s_ + impl_.offset(id_path)) = '.';
1282 1 ++impl_.decoded_[id_path];
1283 1 return true;
1284 }
1285
1286 // remove '/'
1287 4 auto n = impl_.len(id_port);
1288 4 impl_.split(id_port, n + 1);
1289 4 resize_impl(id_port, n, op);
1290 4 --impl_.decoded_[id_path];
1291 4 return true;
1292 }
1293
1294 3 if(! absolute)
1295 {
1296 // already not absolute
1297 1 return true;
1298 }
1299
1300 // add '/'
1301 2 auto n = impl_.len(id_port);
1302 2 auto dest = resize_impl(
1303 2 id_port, n + 1, op) + n;
1304 2 impl_.split(id_port, n);
1305 2 *dest = '/';
1306 2 ++impl_.decoded_[id_path];
1307 2 return true;
1308 52 }
1309
1310 inline
1311 url_base&
1312 298 url_base::
1313 set_path(
1314 core::string_view s)
1315 {
1316 298 op_t op(*this, &s);
1317 298 encoding_opts opt;
1318
1319 //------------------------------------------------
1320 //
1321 // Calculate encoded size
1322 //
1323 // - "/"s are not encoded
1324 // - "%2F"s are not encoded
1325 //
1326 // - reserved path chars are re-encoded
1327 // - colons in first segment might need to be re-encoded
1328 // - the path might need to receive a prefix
1329 298 auto const n = encoded_size(
1330 s, detail::path_chars, opt);
1331 298 std::size_t n_reencode_colons = 0;
1332 298 core::string_view first_seg;
1333 298 if (!has_scheme() &&
1334 337 !has_authority() &&
1335 39 !s.starts_with('/'))
1336 {
1337 // the first segment with unencoded colons would look
1338 // like the scheme
1339 6 first_seg = detail::to_sv(s);
1340 6 std::size_t p = s.find('/');
1341 6 if (p != core::string_view::npos)
1342 2 first_seg = s.substr(0, p);
1343 6 n_reencode_colons = std::count(
1344 12 first_seg.begin(), first_seg.end(), ':');
1345 }
1346 // the authority can only be followed by an empty or relative path
1347 // if we have an authority and the path is a non-empty relative path, we
1348 // add the "/" prefix to make it valid.
1349 bool make_absolute =
1350 298 has_authority() &&
1351 305 !s.starts_with('/') &&
1352 7 !s.empty();
1353 // a path starting with "//" might look like the authority.
1354 // we add a "/." prefix to prevent that
1355 bool add_dot_segment =
1356 592 !make_absolute &&
1357 294 s.starts_with("//");
1358
1359 //------------------------------------------------
1360 //
1361 // Re-encode data
1362 //
1363 596 auto dest = set_path_impl(
1364 298 n + make_absolute + 2 * n_reencode_colons + 2 * add_dot_segment, op);
1365 298 impl_.decoded_[id_path] = 0;
1366 298 if (!dest)
1367 {
1368 3 impl_.nseg_ = 0;
1369 3 return *this;
1370 }
1371 295 if (make_absolute)
1372 {
1373 4 *dest++ = '/';
1374 4 impl_.decoded_[id_path] += 1;
1375 }
1376 291 else if (add_dot_segment)
1377 {
1378 12 *dest++ = '/';
1379 12 *dest++ = '.';
1380 12 impl_.decoded_[id_path] += 2;
1381 }
1382 590 dest += encode_unsafe(
1383 dest,
1384 295 impl_.get(id_query).data() - dest,
1385 first_seg,
1386 295 detail::segment_chars - ':',
1387 opt);
1388 295 dest += encode_unsafe(
1389 dest,
1390 295 impl_.get(id_query).data() - dest,
1391 s.substr(first_seg.size()),
1392 detail::path_chars,
1393 opt);
1394 295 impl_.decoded_[id_path] +=
1395 295 detail::to_size_type(s.size());
1396 295 BOOST_ASSERT(!dest || dest == impl_.get(id_query).data());
1397 295 BOOST_ASSERT(
1398 impl_.decoded_[id_path] ==
1399 s.size() + make_absolute + 2 * add_dot_segment);
1400
1401 //------------------------------------------------
1402 //
1403 // Update path parameters
1404 //
1405 // get the encoded_path with the replacements we applied
1406 295 if (s == "/")
1407 {
1408 // "/" maps to sequence {}
1409 144 impl_.nseg_ = 0;
1410 }
1411 151 else if (!s.empty())
1412 {
1413 146 if (s.starts_with("/./"))
1414 1 s = s.substr(2);
1415 // count segments as number of '/'s + 1
1416 292 impl_.nseg_ = detail::to_size_type(
1417 146 std::count(
1418 292 s.begin() + 1, s.end(), '/') + 1);
1419 }
1420 else
1421 {
1422 // an empty relative path maps to sequence {}
1423 5 impl_.nseg_ = 0;
1424 }
1425
1426 295 check_invariants();
1427 295 return *this;
1428 298 }
1429
1430 inline
1431 url_base&
1432 299 url_base::
1433 set_encoded_path(
1434 pct_string_view s)
1435 {
1436 299 op_t op(*this, &detail::ref(s));
1437
1438 //------------------------------------------------
1439 //
1440 // Calculate re-encoded output size
1441 //
1442 // - reserved path chars are re-encoded
1443 // - colons in first segment might need to be re-encoded
1444 // - the path might need to receive a prefix
1445 299 auto const n = detail::re_encoded_size_unsafe(
1446 s, detail::path_chars);
1447 299 std::size_t n_reencode_colons = 0;
1448 299 core::string_view first_seg;
1449 299 if (!has_scheme() &&
1450 332 !has_authority() &&
1451 33 !s.starts_with('/'))
1452 {
1453 // the first segment with unencoded colons would look
1454 // like the scheme
1455 26 first_seg = detail::to_sv(s);
1456 26 std::size_t p = s.find('/');
1457 26 if (p != core::string_view::npos)
1458 9 first_seg = s.substr(0, p);
1459 26 n_reencode_colons = std::count(
1460 52 first_seg.begin(), first_seg.end(), ':');
1461 }
1462 // the authority can only be followed by an empty or relative path
1463 // if we have an authority and the path is a non-empty relative path, we
1464 // add the "/" prefix to make it valid.
1465 bool make_absolute =
1466 299 has_authority() &&
1467 443 !s.starts_with('/') &&
1468 144 !s.empty();
1469 // a path starting with "//" might look like the authority
1470 // we add a "/." prefix to prevent that
1471 bool add_dot_segment =
1472 502 !make_absolute &&
1473 362 !has_authority() &&
1474 63 s.starts_with("//");
1475
1476 //------------------------------------------------
1477 //
1478 // Re-encode data
1479 //
1480 598 auto dest = set_path_impl(
1481 299 n + make_absolute + 2 * n_reencode_colons + 2 * add_dot_segment, op);
1482 299 impl_.decoded_[id_path] = 0;
1483 299 if (!dest)
1484 {
1485 2 impl_.nseg_ = 0;
1486 2 return *this;
1487 }
1488 297 if (make_absolute)
1489 {
1490 96 *dest++ = '/';
1491 96 impl_.decoded_[id_path] += 1;
1492 }
1493 201 else if (add_dot_segment)
1494 {
1495 5 *dest++ = '/';
1496 5 *dest++ = '.';
1497 5 impl_.decoded_[id_path] += 2;
1498 }
1499 297 impl_.decoded_[id_path] +=
1500 297 detail::to_size_type(detail::re_encode_unsafe(
1501 dest,
1502 297 impl_.get(id_query).data(),
1503 first_seg,
1504 297 detail::segment_chars - ':'));
1505 297 impl_.decoded_[id_path] +=
1506 594 detail::to_size_type(detail::re_encode_unsafe(
1507 dest,
1508 297 impl_.get(id_query).data(),
1509 s.substr(first_seg.size()),
1510 detail::path_chars));
1511 297 BOOST_ASSERT(dest == impl_.get(id_query).data());
1512 297 BOOST_ASSERT(
1513 impl_.decoded_[id_path] ==
1514 s.decoded_size() + make_absolute + 2 * add_dot_segment);
1515
1516 //------------------------------------------------
1517 //
1518 // Update path parameters
1519 //
1520 // get the encoded_path with the replacements we applied
1521 297 if (s == "/")
1522 {
1523 // "/" maps to sequence {}
1524 19 impl_.nseg_ = 0;
1525 }
1526 278 else if (!s.empty())
1527 {
1528 227 if (s.starts_with("/./"))
1529 7 s = s.substr(2);
1530 // count segments as number of '/'s + 1
1531 454 impl_.nseg_ = detail::to_size_type(
1532 227 std::count(
1533 454 s.begin() + 1, s.end(), '/') + 1);
1534 }
1535 else
1536 {
1537 // an empty relative path maps to sequence {}
1538 51 impl_.nseg_ = 0;
1539 }
1540
1541 297 check_invariants();
1542 297 return *this;
1543 299 }
1544
1545 inline
1546 segments_ref
1547 1146 url_base::
1548 segments() noexcept
1549 {
1550 1146 return {*this};
1551 }
1552
1553 inline
1554 segments_encoded_ref
1555 497 url_base::
1556 encoded_segments() noexcept
1557 {
1558 497 return {*this};
1559 }
1560
1561 //------------------------------------------------
1562 //
1563 // Query
1564 //
1565 //------------------------------------------------
1566
1567 inline
1568 url_base&
1569 169 url_base::
1570 set_query(
1571 core::string_view s)
1572 {
1573 169 edit_params(
1574 169 detail::params_iter_impl(impl_),
1575 169 detail::params_iter_impl(impl_, 0),
1576 338 detail::query_string_iter(s, true));
1577 169 return *this;
1578 }
1579
1580 inline
1581 url_base&
1582 179 url_base::
1583 set_encoded_query(
1584 pct_string_view s)
1585 {
1586 179 op_t op(*this);
1587 179 std::size_t n = 0; // encoded size
1588 179 std::size_t nparam = 1; // param count
1589 179 auto const end = s.end();
1590 179 auto p = s.begin();
1591
1592 // measure
1593 742 while(p != end)
1594 {
1595 563 if(*p == '&')
1596 {
1597 4 ++p;
1598 4 ++n;
1599 4 ++nparam;
1600 }
1601 559 else if(*p != '%')
1602 {
1603 533 if(detail::query_chars(*p))
1604 518 n += 1; // allowed
1605 else
1606 15 n += 3; // escaped
1607 533 ++p;
1608 }
1609 else
1610 {
1611 // escape
1612 26 n += 3;
1613 26 p += 3;
1614 }
1615 }
1616
1617 // resize
1618 179 auto dest = resize_impl(
1619 179 id_query, n + 1, op);
1620 179 *dest++ = '?';
1621
1622 // encode
1623 179 impl_.decoded_[id_query] =
1624 358 detail::to_size_type(detail::re_encode_unsafe(
1625 dest,
1626 179 dest + n,
1627 s,
1628 detail::query_chars));
1629 179 BOOST_ASSERT(
1630 impl_.decoded_[id_query] ==
1631 s.decoded_size());
1632 179 impl_.nparam_ =
1633 179 detail::to_size_type(nparam);
1634 179 return *this;
1635 179 }
1636
1637 inline
1638 params_ref
1639 977 url_base::
1640 params() noexcept
1641 {
1642 return params_ref(
1643 *this,
1644 encoding_opts{
1645 977 true, false, false});
1646 }
1647
1648 inline
1649 params_ref
1650 4 url_base::
1651 params(encoding_opts opt) noexcept
1652 {
1653 4 return {*this, opt};
1654 }
1655
1656 inline
1657 params_encoded_ref
1658 77 url_base::
1659 encoded_params() noexcept
1660 {
1661 77 return {*this};
1662 }
1663
1664 inline
1665 url_base&
1666 1 url_base::
1667 set_params(
1668 std::initializer_list<param_view> ps,
1669 encoding_opts opts) noexcept
1670 {
1671 1 params(opts).assign(ps);
1672 1 return *this;
1673 }
1674
1675 inline
1676 url_base&
1677 1 url_base::
1678 set_encoded_params( std::initializer_list< param_pct_view > ps ) noexcept
1679 {
1680 1 encoded_params().assign(ps);
1681 1 return *this;
1682 }
1683
1684 inline
1685 url_base&
1686 990 url_base::
1687 remove_query() noexcept
1688 {
1689 990 op_t op(*this);
1690 990 resize_impl(id_query, 0, op);
1691 990 impl_.nparam_ = 0;
1692 990 impl_.decoded_[id_query] = 0;
1693 1980 return *this;
1694 990 }
1695
1696 //------------------------------------------------
1697 //
1698 // Fragment
1699 //
1700 //------------------------------------------------
1701
1702 inline
1703 url_base&
1704 407 url_base::
1705 remove_fragment() noexcept
1706 {
1707 407 op_t op(*this);
1708 407 resize_impl(id_frag, 0, op);
1709 407 impl_.decoded_[id_frag] = 0;
1710 814 return *this;
1711 407 }
1712
1713 inline
1714 url_base&
1715 151 url_base::
1716 set_fragment(core::string_view s)
1717 {
1718 151 op_t op(*this, &s);
1719 151 encoding_opts opt;
1720 151 auto const n = encoded_size(
1721 s,
1722 detail::fragment_chars,
1723 opt);
1724 151 auto dest = resize_impl(
1725 id_frag, n + 1, op);
1726 151 *dest++ = '#';
1727 151 encode_unsafe(
1728 dest,
1729 n,
1730 s,
1731 detail::fragment_chars,
1732 opt);
1733 151 impl_.decoded_[id_frag] =
1734 151 detail::to_size_type(s.size());
1735 151 return *this;
1736 151 }
1737
1738 inline
1739 url_base&
1740 183 url_base::
1741 set_encoded_fragment(
1742 pct_string_view s)
1743 {
1744 183 op_t op(*this, &detail::ref(s));
1745 auto const n =
1746 183 detail::re_encoded_size_unsafe(
1747 s,
1748 detail::fragment_chars);
1749 183 auto dest = resize_impl(
1750 183 id_frag, n + 1, op);
1751 183 *dest++ = '#';
1752 183 impl_.decoded_[id_frag] =
1753 366 detail::to_size_type(detail::re_encode_unsafe(
1754 dest,
1755 183 dest + n,
1756 s,
1757 detail::fragment_chars));
1758 183 BOOST_ASSERT(
1759 impl_.decoded_[id_frag] ==
1760 s.decoded_size());
1761 183 return *this;
1762 183 }
1763
1764 //------------------------------------------------
1765 //
1766 // Resolution
1767 //
1768 //------------------------------------------------
1769
1770 inline
1771 system::result<void>
1772 516 url_base::
1773 resolve(
1774 url_view_base const& ref)
1775 {
1776 519 if (this == &ref &&
1777 3 has_scheme())
1778 {
1779 2 normalize_path();
1780 2 return {};
1781 }
1782
1783 514 if(! has_scheme())
1784 {
1785 2 BOOST_URL_RETURN_EC(error::not_a_base);
1786 }
1787
1788 512 op_t op(*this);
1789
1790 //
1791 // 5.2.2. Transform References
1792 // https://datatracker.ietf.org/doc/html/rfc3986#section-5.2.2
1793 //
1794
1795 782 if( ref.has_scheme() &&
1796 270 ref.scheme() != scheme())
1797 {
1798 202 reserve_impl(ref.size(), op);
1799 202 copy(ref);
1800 202 normalize_path();
1801 202 return {};
1802 }
1803 310 if(ref.has_authority())
1804 {
1805 78 reserve_impl(
1806 78 impl_.offset(id_user) + ref.size(), op);
1807 78 set_encoded_authority(
1808 ref.encoded_authority());
1809 78 set_encoded_path(
1810 ref.encoded_path());
1811 78 if (ref.encoded_path().empty())
1812 33 set_path_absolute(false);
1813 else
1814 45 normalize_path();
1815 78 if(ref.has_query())
1816 8 set_encoded_query(
1817 ref.encoded_query());
1818 else
1819 70 remove_query();
1820 78 if(ref.has_fragment())
1821 7 set_encoded_fragment(
1822 ref.encoded_fragment());
1823 else
1824 71 remove_fragment();
1825 78 return {};
1826 }
1827 232 if(ref.encoded_path().empty())
1828 {
1829 43 reserve_impl(
1830 43 impl_.offset(id_query) +
1831 43 ref.size(), op);
1832 43 normalize_path();
1833 43 if(ref.has_query())
1834 {
1835 12 set_encoded_query(
1836 ref.encoded_query());
1837 }
1838 43 if(ref.has_fragment())
1839 20 set_encoded_fragment(
1840 ref.encoded_fragment());
1841 else
1842 23 remove_fragment();
1843 43 return {};
1844 }
1845 189 if(ref.is_path_absolute())
1846 {
1847 41 reserve_impl(
1848 41 impl_.offset(id_path) +
1849 41 ref.size(), op);
1850 41 set_encoded_path(
1851 ref.encoded_path());
1852 41 normalize_path();
1853 41 if(ref.has_query())
1854 4 set_encoded_query(
1855 ref.encoded_query());
1856 else
1857 37 remove_query();
1858 41 if(ref.has_fragment())
1859 2 set_encoded_fragment(
1860 ref.encoded_fragment());
1861 else
1862 39 remove_fragment();
1863 41 return {};
1864 }
1865 // General case: ref is relative path
1866 148 reserve_impl(
1867 148 impl_.offset(id_query) +
1868 148 ref.size(), op);
1869 // 5.2.3. Merge Paths
1870 148 auto es = encoded_segments();
1871 148 if(es.size() > 0)
1872 {
1873 141 es.pop_back();
1874 }
1875 296 es.insert(es.end(),
1876 148 ref.encoded_segments().begin(),
1877 148 ref.encoded_segments().end());
1878 148 normalize_path();
1879 148 if(ref.has_query())
1880 10 set_encoded_query(
1881 ref.encoded_query());
1882 else
1883 138 remove_query();
1884 148 if(ref.has_fragment())
1885 13 set_encoded_fragment(
1886 ref.encoded_fragment());
1887 else
1888 135 remove_fragment();
1889 148 return {};
1890 512 }
1891
1892 //------------------------------------------------
1893 //
1894 // Normalization
1895 //
1896 //------------------------------------------------
1897
1898 template <
1899 class AllowedCharset,
1900 class IgnoredCharset>
1901 void
1902 2880 url_base::
1903 normalize_octets_impl(
1904 int id,
1905 AllowedCharset const& allowed,
1906 IgnoredCharset const& ignored,
1907 op_t& op) noexcept
1908 {
1909 2880 char* it = s_ + impl_.offset(id);
1910 2880 char* end = s_ + impl_.offset(id + 1);
1911 2880 char d = 0;
1912 2880 char* dest = it;
1913 16129 while (it < end)
1914 {
1915 13249 if (*it != '%')
1916 {
1917 13031 *dest = *it;
1918 13031 ++it;
1919 13031 ++dest;
1920 13031 continue;
1921 }
1922 218 BOOST_ASSERT(end - it >= 3);
1923
1924 // decode unreserved octets
1925 218 d = detail::decode_one(it + 1);
1926 314 if (allowed(d) &&
1927 96 !ignored(d))
1928 {
1929 87 *dest = d;
1930 87 it += 3;
1931 87 ++dest;
1932 87 continue;
1933 }
1934
1935 // uppercase percent-encoding triplets
1936 131 *dest++ = '%';
1937 131 ++it;
1938 131 *dest++ = grammar::to_upper(*it++);
1939 131 *dest++ = grammar::to_upper(*it++);
1940 }
1941 2880 if (it != dest)
1942 {
1943 35 auto diff = it - dest;
1944 35 auto n = impl_.len(id) - diff;
1945 35 shrink_impl(id, n, op);
1946 35 s_[size()] = '\0';
1947 }
1948 2880 }
1949
1950 template<class CharSet>
1951 void
1952 2681 url_base::
1953 normalize_octets_impl(
1954 int idx,
1955 CharSet const& allowed,
1956 op_t& op) noexcept
1957 {
1958 2681 return normalize_octets_impl(
1959 2681 idx, allowed, detail::empty_chars, op);
1960 }
1961
1962 inline
1963 url_base&
1964 198 url_base::
1965 normalize_scheme()
1966 {
1967 198 to_lower_impl(id_scheme);
1968 198 return *this;
1969 }
1970
1971 inline
1972 url_base&
1973 544 url_base::
1974 normalize_authority()
1975 {
1976 544 op_t op(*this);
1977
1978 // normalize host
1979 544 if (host_type() == urls::host_type::name)
1980 {
1981 351 normalize_octets_impl(
1982 id_host,
1983 detail::reg_name_chars, op);
1984 }
1985 544 decoded_to_lower_impl(id_host);
1986
1987 // normalize password
1988 544 normalize_octets_impl(id_pass, detail::password_chars, op);
1989
1990 // normalize user
1991 544 normalize_octets_impl(id_user, detail::user_chars, op);
1992 1088 return *this;
1993 544 }
1994
1995 inline
1996 url_base&
1997 1046 url_base::
1998 normalize_path()
1999 {
2000 1046 op_t op(*this);
2001 1046 normalize_octets_impl(id_path, detail::segment_chars, op);
2002 1046 core::string_view p = impl_.get(id_path);
2003 1046 char* p_dest = s_ + impl_.offset(id_path);
2004 1046 char* p_end = s_ + impl_.offset(id_path + 1);
2005 1046 auto pn = p.size();
2006 1046 auto skip_dot = 0;
2007 1046 bool encode_colons = false;
2008 1046 core::string_view first_seg;
2009
2010 //------------------------------------------------
2011 //
2012 // Determine unnecessary initial dot segments to skip and
2013 // if we need to encode colons in the first segment
2014 //
2015 1046 if (
2016 1340 !has_authority() &&
2017 294 p.starts_with("/./"))
2018 {
2019 // check if removing the "/./" would result in "//"
2020 // ex: "/.//", "/././/", "/././/", ...
2021 14 skip_dot = 2;
2022 15 while (p.substr(skip_dot, 3).starts_with("/./"))
2023 1 skip_dot += 2;
2024 14 if (p.substr(skip_dot).starts_with("//"))
2025 11 skip_dot = 2;
2026 else
2027 3 skip_dot = 0;
2028 }
2029 1032 else if (
2030 1092 !has_scheme() &&
2031 60 !has_authority())
2032 {
2033 41 if (p.starts_with("./"))
2034 {
2035 // check if removing the "./" would result in "//"
2036 // ex: ".//", "././/", "././/", ...
2037 7 skip_dot = 1;
2038 10 while (p.substr(skip_dot, 3).starts_with("/./"))
2039 3 skip_dot += 2;
2040 7 if (p.substr(skip_dot).starts_with("//"))
2041 2 skip_dot = 2;
2042 else
2043 5 skip_dot = 0;
2044
2045 7 if ( !skip_dot )
2046 {
2047 // check if removing "./"s would leave us
2048 // a first segment with an ambiguous ":"
2049 5 first_seg = p.substr(2);
2050 7 while (first_seg.starts_with("./"))
2051 2 first_seg = first_seg.substr(2);
2052 5 auto i = first_seg.find('/');
2053 5 if (i != core::string_view::npos)
2054 1 first_seg = first_seg.substr(0, i);
2055 5 encode_colons = first_seg.contains(':');
2056 }
2057 }
2058 else
2059 {
2060 // check if normalize_octets_impl
2061 // didn't already create a ":"
2062 // in the first segment
2063 34 first_seg = p;
2064 34 auto i = first_seg.find('/');
2065 34 if (i != core::string_view::npos)
2066 17 first_seg = p.substr(0, i);
2067 34 encode_colons = first_seg.contains(':');
2068 }
2069 }
2070
2071 //------------------------------------------------
2072 //
2073 // Encode colons in the first segment
2074 //
2075 1046 if (encode_colons)
2076 {
2077 // prepend with "./"
2078 // (resize_impl never throws)
2079 auto cn =
2080 5 std::count(
2081 first_seg.begin(),
2082 first_seg.end(),
2083 5 ':');
2084 5 resize_impl(
2085 5 id_path, pn + (2 * cn), op);
2086 // move the 2nd, 3rd, ... segments
2087 5 auto begin = s_ + impl_.offset(id_path);
2088 5 auto it = begin;
2089 5 auto end = begin + pn;
2090 11 while (core::string_view(it, 2) == "./")
2091 6 it += 2;
2092 57 while (it != end &&
2093 52 *it != '/')
2094 52 ++it;
2095 // we don't need op here because this is
2096 // an internal operation
2097 5 std::memmove(it + (2 * cn), it, end - it);
2098
2099 // move 1st segment
2100 5 auto src = s_ + impl_.offset(id_path) + pn;
2101 5 auto dest = s_ + impl_.offset(id_query);
2102 5 src -= end - it;
2103 5 dest -= end - it;
2104 5 pn -= end - it;
2105 do {
2106 64 --src;
2107 64 --dest;
2108 64 if (*src != ':')
2109 {
2110 57 *dest = *src;
2111 }
2112 else
2113 {
2114 // use uppercase as required by
2115 // syntax-based normalization
2116 7 *dest-- = 'A';
2117 7 *dest-- = '3';
2118 7 *dest = '%';
2119 }
2120 64 --pn;
2121 64 } while (pn);
2122 5 skip_dot = 0;
2123 5 p = impl_.get(id_path);
2124 5 pn = p.size();
2125 5 p_dest = s_ + impl_.offset(id_path);
2126 5 p_end = s_ + impl_.offset(id_path + 1);
2127 }
2128
2129 //------------------------------------------------
2130 //
2131 // Remove "." and ".." segments
2132 //
2133 1046 p.remove_prefix(skip_dot);
2134 1046 p_dest += skip_dot;
2135 1046 auto n = detail::remove_dot_segments(
2136 p_dest, p_end, p);
2137
2138 //------------------------------------------------
2139 //
2140 // Update path parameters
2141 //
2142 1046 if (n != pn)
2143 {
2144 140 BOOST_ASSERT(n < pn);
2145 140 shrink_impl(id_path, n + skip_dot, op);
2146 140 p = encoded_path();
2147 140 if (p == "/")
2148 9 impl_.nseg_ = 0;
2149 131 else if (!p.empty())
2150 258 impl_.nseg_ = detail::to_size_type(
2151 129 std::count(
2152 258 p.begin() + 1, p.end(), '/') + 1);
2153 else
2154 2 impl_.nseg_ = 0;
2155 140 impl_.decoded_[id_path] =
2156 140 detail::to_size_type(detail::decode_bytes_unsafe(
2157 impl_.get(id_path)));
2158 }
2159 1046 return *this;
2160 1046 }
2161
2162 inline
2163 url_base&
2164 199 url_base::
2165 normalize_query()
2166 {
2167 199 op_t op(*this);
2168 199 normalize_octets_impl(
2169 id_query,
2170 detail::query_chars,
2171 detail::query_ignore_chars,
2172 op);
2173 398 return *this;
2174 199 }
2175
2176 inline
2177 url_base&
2178 196 url_base::
2179 normalize_fragment()
2180 {
2181 196 op_t op(*this);
2182 196 normalize_octets_impl(
2183 id_frag, detail::fragment_chars, op);
2184 392 return *this;
2185 196 }
2186
2187 inline
2188 url_base&
2189 195 url_base::
2190 normalize()
2191 {
2192 195 normalize_fragment();
2193 195 normalize_query();
2194 195 normalize_path();
2195 195 normalize_authority();
2196 195 normalize_scheme();
2197 195 return *this;
2198 }
2199
2200 //------------------------------------------------
2201 //
2202 // Implementation
2203 //
2204 //------------------------------------------------
2205
2206 inline
2207 void
2208 43324 url_base::
2209 check_invariants() const noexcept
2210 {
2211 43324 BOOST_ASSERT(
2212 impl_.len(id_scheme) == 0 ||
2213 impl_.get(id_scheme).ends_with(':'));
2214 43324 BOOST_ASSERT(
2215 impl_.len(id_user) == 0 ||
2216 impl_.get(id_user).starts_with("//"));
2217 43324 BOOST_ASSERT(
2218 impl_.len(id_pass) == 0 ||
2219 impl_.get(id_user).starts_with("//"));
2220 43324 BOOST_ASSERT(
2221 impl_.len(id_pass) == 0 ||
2222 (impl_.len(id_pass) == 1 &&
2223 impl_.get(id_pass) == "@") ||
2224 (impl_.len(id_pass) > 1 &&
2225 impl_.get(id_pass).starts_with(':') &&
2226 impl_.get(id_pass).ends_with('@')));
2227 43324 BOOST_ASSERT(
2228 impl_.len(id_user, id_path) == 0 ||
2229 impl_.get(id_user).starts_with("//"));
2230 43324 BOOST_ASSERT(impl_.decoded_[id_path] >=
2231 ((impl_.len(id_path) + 2) / 3));
2232 43324 BOOST_ASSERT(
2233 impl_.len(id_port) == 0 ||
2234 impl_.get(id_port).starts_with(':'));
2235 43324 BOOST_ASSERT(
2236 impl_.len(id_query) == 0 ||
2237 impl_.get(id_query).starts_with('?'));
2238 43324 BOOST_ASSERT(
2239 (impl_.len(id_query) == 0 && impl_.nparam_ == 0) ||
2240 (impl_.len(id_query) > 0 && impl_.nparam_ > 0));
2241 43324 BOOST_ASSERT(
2242 impl_.len(id_frag) == 0 ||
2243 impl_.get(id_frag).starts_with('#'));
2244 43324 BOOST_ASSERT(c_str()[size()] == '\0');
2245 43324 }
2246
2247 inline
2248 char*
2249 5492 url_base::
2250 resize_impl(
2251 int id,
2252 std::size_t new_size,
2253 op_t& op)
2254 {
2255 5492 return resize_impl(
2256 5492 id, id + 1, new_size, op);
2257 }
2258
2259 inline
2260 char*
2261 6217 url_base::
2262 resize_impl(
2263 int first,
2264 int last,
2265 std::size_t new_len,
2266 op_t& op)
2267 {
2268 6217 auto const n0 = impl_.len(first, last);
2269 6217 if(new_len == 0 && n0 == 0)
2270 1090 return s_ + impl_.offset(first);
2271 5127 if(new_len <= n0)
2272 2193 return shrink_impl(
2273 2193 first, last, new_len, op);
2274
2275 // growing
2276 2934 std::size_t n = new_len - n0;
2277 2934 reserve_impl(size() + n, op);
2278 auto const pos =
2279 2934 impl_.offset(last);
2280 // adjust chars
2281 2934 op.move(
2282 2934 s_ + pos + n,
2283 2934 s_ + pos,
2284 2934 impl_.offset(id_end) -
2285 pos + 1);
2286 // collapse (first, last)
2287 2934 impl_.collapse(first, last,
2288 2934 impl_.offset(last) + n);
2289 // shift (last, end) right
2290 2934 impl_.adjust_right(last, id_end, n);
2291 2934 s_[size()] = '\0';
2292 2934 return s_ + impl_.offset(first);
2293 }
2294
2295 inline
2296 char*
2297 175 url_base::
2298 shrink_impl(
2299 int id,
2300 std::size_t new_size,
2301 op_t& op)
2302 {
2303 175 return shrink_impl(
2304 175 id, id + 1, new_size, op);
2305 }
2306
2307 inline
2308 char*
2309 2368 url_base::
2310 shrink_impl(
2311 int first,
2312 int last,
2313 std::size_t new_len,
2314 op_t& op)
2315 {
2316 // shrinking
2317 2368 auto const n0 = impl_.len(first, last);
2318 2368 BOOST_ASSERT(new_len <= n0);
2319 2368 std::size_t n = n0 - new_len;
2320 auto const pos =
2321 2368 impl_.offset(last);
2322 // adjust chars
2323 2368 op.move(
2324 2368 s_ + pos - n,
2325 2368 s_ + pos,
2326 2368 impl_.offset(
2327 2368 id_end) - pos + 1);
2328 // collapse (first, last)
2329 2368 impl_.collapse(first, last,
2330 2368 impl_.offset(last) - n);
2331 // shift (last, end) left
2332 2368 impl_.adjust_left(last, id_end, n);
2333 2368 s_[size()] = '\0';
2334 2368 return s_ + impl_.offset(first);
2335 }
2336
2337 //------------------------------------------------
2338
2339 inline
2340 void
2341 526 url_base::
2342 set_scheme_impl(
2343 core::string_view s,
2344 urls::scheme id)
2345 {
2346 526 op_t op(*this, &s);
2347 526 check_invariants();
2348 574 grammar::parse(
2349 48 s, detail::scheme_rule()
2350 574 ).value(BOOST_URL_POS);
2351 478 auto const n = s.size();
2352 478 auto const p = impl_.offset(id_path);
2353
2354 // check for "./" prefix
2355 bool const has_dot =
2356 1434 [this, p]
2357 {
2358 478 if(impl_.nseg_ == 0)
2359 213 return false;
2360 265 if(first_segment().size() < 2)
2361 42 return false;
2362 223 auto const src = s_ + p;
2363 223 if(src[0] != '.')
2364 220 return false;
2365 3 if(src[1] != '/')
2366 return false;
2367 3 return true;
2368 478 }();
2369
2370 // Remove "./"
2371 478 if(has_dot)
2372 {
2373 // do this first, for
2374 // strong exception safety
2375 6 reserve_impl(
2376 3 size() + n + 1 - 2, op);
2377 3 op.move(
2378 3 s_ + p,
2379 3 s_ + p + 2,
2380 3 size() + 1 -
2381 (p + 2));
2382 3 impl_.set_size(
2383 id_path,
2384 3 impl_.len(id_path) - 2);
2385 3 s_[size()] = '\0';
2386 }
2387
2388 478 auto dest = resize_impl(
2389 id_scheme, n + 1, op);
2390 478 s.copy(dest, n);
2391 478 dest[n] = ':';
2392 478 impl_.scheme_ = id;
2393 478 check_invariants();
2394 526 }
2395
2396 inline
2397 char*
2398 396 url_base::
2399 set_user_impl(
2400 std::size_t n,
2401 op_t& op)
2402 {
2403 396 check_invariants();
2404 396 if(impl_.len(id_pass) != 0)
2405 {
2406 // keep "//"
2407 147 auto dest = resize_impl(
2408 id_user, 2 + n, op);
2409 147 check_invariants();
2410 147 return dest + 2;
2411 }
2412 // add authority
2413 bool const make_absolute =
2414 333 !is_path_absolute() &&
2415 84 !impl_.get(id_path).empty();
2416 498 auto dest = resize_impl(
2417 249 id_user, 2 + n + 1 + make_absolute, op);
2418 249 impl_.split(id_user, 2 + n);
2419 249 dest[0] = '/';
2420 249 dest[1] = '/';
2421 249 dest[2 + n] = '@';
2422 249 if (make_absolute)
2423 {
2424 6 impl_.split(id_pass, 1);
2425 6 impl_.split(id_host, 0);
2426 6 impl_.split(id_port, 0);
2427 6 dest[3 + n] = '/';
2428 }
2429 249 check_invariants();
2430 249 return dest + 2;
2431 }
2432
2433 inline
2434 char*
2435 377 url_base::
2436 set_password_impl(
2437 std::size_t n,
2438 op_t& op)
2439 {
2440 377 check_invariants();
2441 377 if(impl_.len(id_user) != 0)
2442 {
2443 // already have authority
2444 313 auto const dest = resize_impl(
2445 id_pass, 1 + n + 1, op);
2446 313 dest[0] = ':';
2447 313 dest[n + 1] = '@';
2448 313 check_invariants();
2449 313 return dest + 1;
2450 }
2451 // add authority
2452 bool const make_absolute =
2453 104 !is_path_absolute() &&
2454 40 !impl_.get(id_path).empty();
2455 auto const dest =
2456 128 resize_impl(
2457 id_user, id_host,
2458 64 2 + 1 + n + 1 + make_absolute, op);
2459 64 impl_.split(id_user, 2);
2460 64 dest[0] = '/';
2461 64 dest[1] = '/';
2462 64 dest[2] = ':';
2463 64 dest[2 + n + 1] = '@';
2464 64 if (make_absolute)
2465 {
2466 7 impl_.split(id_pass, 2 + n);
2467 7 impl_.split(id_host, 0);
2468 7 impl_.split(id_port, 0);
2469 7 dest[4 + n] = '/';
2470 }
2471 64 check_invariants();
2472 64 return dest + 3;
2473 }
2474
2475 inline
2476 char*
2477 277 url_base::
2478 set_userinfo_impl(
2479 std::size_t n,
2480 op_t& op)
2481 {
2482 // "//" {dest} "@"
2483 277 check_invariants();
2484 bool const make_absolute =
2485 406 !is_path_absolute() &&
2486 129 !impl_.get(id_path).empty();
2487 554 auto dest = resize_impl(
2488 277 id_user, id_host, n + 3 + make_absolute, op);
2489 277 impl_.split(id_user, n + 2);
2490 277 dest[0] = '/';
2491 277 dest[1] = '/';
2492 277 dest[n + 2] = '@';
2493 277 if (make_absolute)
2494 {
2495 3 impl_.split(id_pass, 1);
2496 3 impl_.split(id_host, 0);
2497 3 impl_.split(id_port, 0);
2498 3 dest[3 + n] = '/';
2499 }
2500 277 check_invariants();
2501 277 return dest + 2;
2502 }
2503
2504 inline
2505 char*
2506 721 url_base::
2507 set_host_impl(
2508 std::size_t n,
2509 op_t& op)
2510 {
2511 721 check_invariants();
2512 721 if(impl_.len(id_user) == 0)
2513 {
2514 // add authority
2515 bool make_absolute =
2516 320 !is_path_absolute() &&
2517 147 impl_.len(id_path) != 0;
2518 173 auto pn = impl_.len(id_path);
2519 346 auto dest = resize_impl(
2520 173 id_user, n + 2 + make_absolute, op);
2521 173 impl_.split(id_user, 2);
2522 173 impl_.split(id_pass, 0);
2523 173 impl_.split(id_host, n);
2524 173 impl_.split(id_port, 0);
2525 173 impl_.split(id_path, pn + make_absolute);
2526 173 if (make_absolute)
2527 {
2528 9 dest[n + 2] = '/';
2529 9 ++impl_.decoded_[id_path];
2530 }
2531 173 dest[0] = '/';
2532 173 dest[1] = '/';
2533 173 check_invariants();
2534 173 return dest + 2;
2535 }
2536 // already have authority
2537 548 auto const dest = resize_impl(
2538 id_host, n, op);
2539 548 check_invariants();
2540 548 return dest;
2541 }
2542
2543 inline
2544 char*
2545 604 url_base::
2546 set_port_impl(
2547 std::size_t n,
2548 op_t& op)
2549 {
2550 604 check_invariants();
2551 604 if(impl_.len(id_user) != 0)
2552 {
2553 // authority exists
2554 503 auto dest = resize_impl(
2555 id_port, n + 1, op);
2556 503 dest[0] = ':';
2557 503 check_invariants();
2558 503 return dest + 1;
2559 }
2560 bool make_absolute =
2561 174 !is_path_absolute() &&
2562 73 impl_.len(id_path) != 0;
2563 202 auto dest = resize_impl(
2564 101 id_user, 3 + n + make_absolute, op);
2565 101 impl_.split(id_user, 2);
2566 101 impl_.split(id_pass, 0);
2567 101 impl_.split(id_host, 0);
2568 101 dest[0] = '/';
2569 101 dest[1] = '/';
2570 101 dest[2] = ':';
2571 101 if (make_absolute)
2572 {
2573 8 impl_.split(id_port, n + 1);
2574 8 dest[n + 3] = '/';
2575 8 ++impl_.decoded_[id_path];
2576 }
2577 101 check_invariants();
2578 101 return dest + 3;
2579 }
2580
2581 inline
2582 char*
2583 597 url_base::
2584 set_path_impl(
2585 std::size_t n,
2586 op_t& op)
2587 {
2588 597 check_invariants();
2589 597 auto const dest = resize_impl(
2590 id_path, n, op);
2591 597 return dest;
2592 }
2593
2594
2595 //------------------------------------------------
2596
2597 // return the first segment of the path.
2598 // this is needed for some algorithms.
2599 inline
2600 core::string_view
2601 303 url_base::
2602 first_segment() const noexcept
2603 {
2604 303 if(impl_.nseg_ == 0)
2605 9 return {};
2606 294 auto const p0 = impl_.cs_ +
2607 294 impl_.offset(id_path) +
2608 294 detail::path_prefix(
2609 294 impl_.get(id_path));
2610 294 auto const end = impl_.cs_ +
2611 294 impl_.offset(id_query);
2612 294 if(impl_.nseg_ == 1)
2613 182 return core::string_view(
2614 91 p0, end - p0);
2615 203 auto p = p0;
2616 910 while(*p != '/')
2617 {
2618 707 BOOST_ASSERT(p < end);
2619 707 ++p;
2620 }
2621 203 return core::string_view(p0, p - p0);
2622 }
2623
2624 inline
2625 detail::segments_iter_impl
2626 2315 url_base::
2627 edit_segments(
2628 detail::segments_iter_impl const& it0,
2629 detail::segments_iter_impl const& it1,
2630 detail::any_segments_iter&& src,
2631 // -1 = preserve
2632 // 0 = make relative (can fail)
2633 // 1 = make absolute
2634 int absolute)
2635 {
2636 // Iterator doesn't belong to this url
2637 2315 BOOST_ASSERT(it0.ref.alias_of(impl_));
2638
2639 // Iterator doesn't belong to this url
2640 2315 BOOST_ASSERT(it1.ref.alias_of(impl_));
2641
2642 // Iterator is in the wrong order
2643 2315 BOOST_ASSERT(it0.index <= it1.index);
2644
2645 // Iterator is out of range
2646 2315 BOOST_ASSERT(it0.index <= impl_.nseg_);
2647 2315 BOOST_ASSERT(it0.pos <= impl_.len(id_path));
2648
2649 // Iterator is out of range
2650 2315 BOOST_ASSERT(it1.index <= impl_.nseg_);
2651 2315 BOOST_ASSERT(it1.pos <= impl_.len(id_path));
2652
2653 //------------------------------------------------
2654 //
2655 // Calculate output prefix
2656 //
2657 // 0 = ""
2658 // 1 = "/"
2659 // 2 = "./"
2660 // 3 = "/./"
2661 //
2662 2315 bool const is_abs = is_path_absolute();
2663 2315 if(has_authority())
2664 {
2665 // Check if the new
2666 // path would be empty
2667 1855 if( src.fast_nseg == 0 &&
2668 911 it0.index == 0 &&
2669 655 it1.index == impl_.nseg_)
2670 {
2671 // VFALCO we don't have
2672 // access to nchar this early
2673 //
2674 //BOOST_ASSERT(nchar == 0);
2675 627 absolute = 0;
2676 }
2677 else
2678 {
2679 // prefix "/" required
2680 1228 absolute = 1;
2681 }
2682 }
2683 460 else if(absolute < 0)
2684 {
2685 460 absolute = is_abs; // preserve
2686 }
2687 2315 auto const path_pos = impl_.offset(id_path);
2688
2689 2315 std::size_t nchar = 0;
2690 2315 std::size_t prefix = 0;
2691 2315 bool encode_colons = false;
2692 2315 bool cp_src_prefix = false;
2693 2315 if(it0.index > 0)
2694 {
2695 // first segment unchanged
2696 870 prefix = src.fast_nseg > 0;
2697 }
2698 1445 else if(src.fast_nseg > 0)
2699 {
2700 // first segment from src
2701 718 if(! src.front.empty())
2702 {
2703 640 if( src.front == "." &&
2704 7 src.fast_nseg > 1)
2705 4 if (src.s.empty())
2706 {
2707 // if front is ".", we need the extra "." in the prefix
2708 // which will maintain the invariant that segments represent
2709 // {"."}
2710 4 prefix = 2 + absolute;
2711 }
2712 else
2713 {
2714 // if the "." prefix is explicitly required from set_path
2715 // we do not include an extra "." segment
2716 prefix = absolute;
2717 cp_src_prefix = true;
2718 }
2719 629 else if(absolute)
2720 542 prefix = 1;
2721 166 else if(has_scheme() ||
2722 79 ! src.front.contains(':'))
2723 82 prefix = 0;
2724 else
2725 {
2726 5 prefix = 0;
2727 5 encode_colons = true;
2728 }
2729 }
2730 else
2731 {
2732 85 prefix = 2 + absolute;
2733 }
2734 }
2735 else
2736 {
2737 // first segment from it1
2738 727 auto const p =
2739 727 impl_.cs_ + path_pos + it1.pos;
2740 1454 switch(impl_.cs_ +
2741 727 impl_.offset(id_query) - p)
2742 {
2743 680 case 0:
2744 // points to end
2745 680 prefix = absolute;
2746 680 break;
2747 36 default:
2748 36 BOOST_ASSERT(*p == '/');
2749 36 if(p[1] != '/')
2750 {
2751 35 if(absolute)
2752 29 prefix = 1;
2753 11 else if(has_scheme() ||
2754 11 ! it1.dereference().contains(':'))
2755 5 prefix = 0;
2756 else
2757 1 prefix = 2;
2758 35 break;
2759 }
2760 // empty
2761 BOOST_FALLTHROUGH;
2762 case 1:
2763 // empty
2764 12 BOOST_ASSERT(*p == '/');
2765 12 prefix = 2 + absolute;
2766 12 break;
2767 }
2768 }
2769
2770 // append '/' to new segs
2771 // if inserting at front.
2772 2315 std::size_t const suffix =
2773 3234 it1.index == 0 &&
2774 2494 impl_.nseg_ > 0 &&
2775 179 src.fast_nseg > 0;
2776
2777 //------------------------------------------------
2778 //
2779 // Measure the number of encoded characters
2780 // of output, and the number of inserted
2781 // segments including internal separators.
2782 //
2783 2315 src.encode_colons = encode_colons;
2784 2315 std::size_t nseg = 0;
2785 2315 if(src.measure(nchar))
2786 {
2787 1279 src.encode_colons = false;
2788 for(;;)
2789 {
2790 1605 ++nseg;
2791 1605 if(! src.measure(nchar))
2792 1277 break;
2793 326 ++nchar;
2794 }
2795 }
2796
2797 2313 switch(src.fast_nseg)
2798 {
2799 1036 case 0:
2800 1036 BOOST_ASSERT(nseg == 0);
2801 1036 break;
2802 1089 case 1:
2803 1089 BOOST_ASSERT(nseg == 1);
2804 1089 break;
2805 188 case 2:
2806 188 BOOST_ASSERT(nseg >= 2);
2807 188 break;
2808 }
2809
2810 //------------------------------------------------
2811 //
2812 // Calculate [pos0, pos1) to remove
2813 //
2814 2313 auto pos0 = it0.pos;
2815 2313 if(it0.index == 0)
2816 {
2817 // patch pos for prefix
2818 1443 pos0 = 0;
2819 }
2820 2313 auto pos1 = it1.pos;
2821 2313 if(it1.index == 0)
2822 {
2823 // patch pos for prefix
2824 919 pos1 = detail::path_prefix(
2825 impl_.get(id_path));
2826 }
2827 1394 else if(
2828 1394 it0.index == 0 &&
2829 524 it1.index < impl_.nseg_ &&
2830 nseg == 0)
2831 {
2832 // Remove the slash from segment it1
2833 // if it is becoming the new first
2834 // segment.
2835 47 ++pos1;
2836 }
2837 // calc decoded size of old range
2838 auto const dn0 =
2839 2313 detail::decode_bytes_unsafe(
2840 core::string_view(
2841 2313 impl_.cs_ +
2842 2313 impl_.offset(id_path) +
2843 pos0,
2844 pos1 - pos0));
2845
2846 //------------------------------------------------
2847 //
2848 // Resize
2849 //
2850 4626 op_t op(*this, &src.s);
2851 char* dest;
2852 char const* end;
2853 {
2854 2313 auto const nremove = pos1 - pos0;
2855 // check overflow
2856 4626 if( nchar <= max_size() && (
2857 2313 prefix + suffix <=
2858 2313 max_size() - nchar))
2859 {
2860 2313 nchar = prefix + nchar + suffix;
2861 3484 if( nchar <= nremove ||
2862 1171 nchar - nremove <=
2863 1171 max_size() - size())
2864 2313 goto ok;
2865 }
2866 // too large
2867 detail::throw_length_error();
2868 2313 ok:
2869 auto const new_size =
2870 2313 size() + nchar - nremove;
2871 2313 reserve_impl(new_size, op);
2872 2313 dest = s_ + path_pos + pos0;
2873 2313 op.move(
2874 2313 dest + nchar,
2875 2313 s_ + path_pos + pos1,
2876 2313 size() - path_pos - pos1);
2877 4626 impl_.set_size(
2878 id_path,
2879 2313 impl_.len(id_path) + nchar - nremove);
2880 2313 BOOST_ASSERT(size() == new_size);
2881 2313 end = dest + nchar;
2882 2313 auto const nseg1 =
2883 2313 static_cast<std::ptrdiff_t>(impl_.nseg_) +
2884 2313 static_cast<std::ptrdiff_t>(nseg) -
2885 2313 static_cast<std::ptrdiff_t>(it1.index) +
2886 2313 static_cast<std::ptrdiff_t>(it0.index) -
2887 static_cast<std::ptrdiff_t>(cp_src_prefix);
2888 2313 BOOST_ASSERT(nseg1 >= 0);
2889 2313 impl_.nseg_ = detail::to_size_type(nseg1);
2890 2313 if(s_)
2891 2306 s_[size()] = '\0';
2892 }
2893
2894 //------------------------------------------------
2895 //
2896 // Output segments and internal separators:
2897 //
2898 // prefix [ segment [ '/' segment ] ] suffix
2899 //
2900 2313 auto const dest0 = dest;
2901 2313 switch(prefix)
2902 {
2903 60 case 3:
2904 60 *dest++ = '/';
2905 60 *dest++ = '.';
2906 60 *dest++ = '/';
2907 60 break;
2908 42 case 2:
2909 42 *dest++ = '.';
2910 BOOST_FALLTHROUGH;
2911 1193 case 1:
2912 1193 *dest++ = '/';
2913 1193 break;
2914 1060 default:
2915 1060 break;
2916 }
2917 2313 src.rewind();
2918 2313 if(nseg > 0)
2919 {
2920 1277 src.encode_colons = encode_colons;
2921 for(;;)
2922 {
2923 1603 src.copy(dest, end);
2924 1603 if(--nseg == 0)
2925 1277 break;
2926 326 *dest++ = '/';
2927 326 src.encode_colons = false;
2928 }
2929 1277 if(suffix)
2930 179 *dest++ = '/';
2931 }
2932 2313 BOOST_ASSERT(dest == dest0 + nchar);
2933
2934 // calc decoded size of new range,
2935 auto const dn =
2936 2313 detail::decode_bytes_unsafe(
2937 2313 core::string_view(dest0, dest - dest0));
2938 2313 if(dn >= dn0)
2939 1401 impl_.decoded_[id_path] +=
2940 1401 detail::to_size_type(dn - dn0);
2941 else
2942 912 impl_.decoded_[id_path] -=
2943 912 detail::to_size_type(dn0 - dn);
2944
2945 return detail::segments_iter_impl(
2946 4626 impl_, pos0, it0.index);
2947 }
2948
2949 //------------------------------------------------
2950
2951 inline
2952 auto
2953 1695 url_base::
2954 edit_params(
2955 detail::params_iter_impl const& it0,
2956 detail::params_iter_impl const& it1,
2957 detail::any_params_iter&& src) ->
2958 detail::params_iter_impl
2959 {
2960 1695 auto pos0 = impl_.offset(id_query);
2961 1695 auto pos1 = pos0 + it1.pos;
2962 1695 pos0 = pos0 + it0.pos;
2963
2964 // Iterators belong to this url
2965 1695 BOOST_ASSERT(it0.ref.alias_of(impl_));
2966 1695 BOOST_ASSERT(it1.ref.alias_of(impl_));
2967
2968 // Iterators is in the right order
2969 1695 BOOST_ASSERT(it0.index <= it1.index);
2970
2971 // Iterators are within range
2972 1695 BOOST_ASSERT(it0.index <= impl_.nparam_);
2973 1695 BOOST_ASSERT(pos0 <= impl_.offset(id_frag));
2974 1695 BOOST_ASSERT(it1.index <= impl_.nparam_);
2975 1695 BOOST_ASSERT(pos1 <= impl_.offset(id_frag));
2976
2977 // calc decoded size of old range,
2978 // minus one if '?' or '&' prefixed
2979 auto dn0 =
2980 static_cast<std::ptrdiff_t>(
2981 1695 detail::decode_bytes_unsafe(
2982 core::string_view(
2983 1695 impl_.cs_ + pos0,
2984 1695 pos1 - pos0)));
2985 1695 if(impl_.len(id_query) > 0)
2986 1157 dn0 -= 1;
2987 1695 if(dn0 < 0)
2988 552 dn0 = 0;
2989
2990 //------------------------------------------------
2991 //
2992 // Measure the number of encoded characters
2993 // of output, and the number of inserted
2994 // segments including internal separators.
2995 //
2996
2997 1695 std::size_t nchar = 0;
2998 1695 std::size_t nparam = 0;
2999 1695 if(src.measure(nchar))
3000 {
3001 1430 ++nchar; // for '?' or '&'
3002 for(;;)
3003 {
3004 1652 ++nparam;
3005 1652 if(! src.measure(nchar))
3006 1430 break;
3007 222 ++nchar; // for '&'
3008 }
3009 }
3010
3011 //------------------------------------------------
3012 //
3013 // Resize
3014 //
3015 1690 op_t op(*this, &src.s0, &src.s1);
3016 char* dest;
3017 char const* end;
3018 {
3019 1690 auto const nremove = pos1 - pos0;
3020 // check overflow
3021 2995 if( nchar > nremove &&
3022 1305 nchar - nremove >
3023 1305 max_size() - size())
3024 {
3025 // too large
3026 detail::throw_length_error();
3027 }
3028 1690 auto const nparam1 =
3029 1690 static_cast<std::ptrdiff_t>(impl_.nparam_) +
3030 1690 static_cast<std::ptrdiff_t>(nparam) -
3031 1690 static_cast<std::ptrdiff_t>(it1.index) +
3032 1690 static_cast<std::ptrdiff_t>(it0.index);
3033 1690 BOOST_ASSERT(nparam1 >= 0);
3034 1690 reserve_impl(size() + nchar - nremove, op);
3035 1690 dest = s_ + pos0;
3036 1690 end = dest + nchar;
3037 1690 if(impl_.nparam_ > 0)
3038 {
3039 // needed when we move
3040 // the beginning of the query
3041 1152 s_[impl_.offset(id_query)] = '&';
3042 }
3043 1690 op.move(
3044 1690 dest + nchar,
3045 1690 impl_.cs_ + pos1,
3046 1690 size() - pos1);
3047 3380 impl_.set_size(
3048 id_query,
3049 1690 impl_.len(id_query) +
3050 nchar - nremove);
3051 1690 impl_.nparam_ =
3052 1690 detail::to_size_type(nparam1);
3053 1690 if(nparam1 > 0)
3054 {
3055 // needed when we erase
3056 // the beginning of the query
3057 1587 s_[impl_.offset(id_query)] = '?';
3058 }
3059 1690 if(s_)
3060 1690 s_[size()] = '\0';
3061 }
3062 1690 auto const dest0 = dest;
3063
3064 //------------------------------------------------
3065 //
3066 // Output params and internal separators:
3067 //
3068 // [ '?' param ] [ '&' param ]
3069 //
3070 1690 if(nparam > 0)
3071 {
3072 1430 if(it0.index == 0)
3073 887 *dest++ = '?';
3074 else
3075 543 *dest++ = '&';
3076 1430 src.rewind();
3077 for(;;)
3078 {
3079 1652 src.copy(dest, end);
3080 1652 if(--nparam == 0)
3081 1430 break;
3082 222 *dest++ = '&';
3083 }
3084 }
3085
3086 // calc decoded size of new range,
3087 // minus one if '?' or '&' prefixed
3088 auto dn =
3089 static_cast<std::ptrdiff_t>(
3090 1690 detail::decode_bytes_unsafe(
3091 1690 core::string_view(dest0, dest - dest0)));
3092 1690 if(impl_.len(id_query) > 0)
3093 1587 dn -= 1;
3094 1690 if(dn < 0)
3095 157 dn = 0;
3096
3097 1690 if(dn >= dn0)
3098 1310 impl_.decoded_[id_query] +=
3099 1310 detail::to_size_type(dn - dn0);
3100 else
3101 380 impl_.decoded_[id_query] -=
3102 380 detail::to_size_type(dn0 - dn);
3103
3104 return detail::params_iter_impl(
3105 1690 impl_,
3106 1690 pos0 - impl_.offset_[id_query],
3107 3380 it0.index);
3108 1690 }
3109
3110 //------------------------------------------------
3111
3112 inline
3113 void
3114 544 url_base::
3115 decoded_to_lower_impl(int id) noexcept
3116 {
3117 544 char* it = s_ + impl_.offset(id);
3118 544 char const* const end = s_ + impl_.offset(id + 1);
3119 3540 while(it < end)
3120 {
3121 2996 if (*it != '%')
3122 {
3123 5962 *it = grammar::to_lower(
3124 2981 *it);
3125 2981 ++it;
3126 2981 continue;
3127 }
3128 15 it += 3;
3129 }
3130 544 }
3131
3132 inline
3133 void
3134 198 url_base::
3135 to_lower_impl(int id) noexcept
3136 {
3137 198 char* it = s_ + impl_.offset(id);
3138 198 char const* const end = s_ + impl_.offset(id + 1);
3139 1027 while(it < end)
3140 {
3141 1658 *it = grammar::to_lower(
3142 829 *it);
3143 829 ++it;
3144 }
3145 198 }
3146
3147 } // urls
3148 } // boost
3149
3150 #endif
3151