. /include/boost/url /impl /url_base.hpp
99.41% Lines (1529/1538) 100.00% Functions (80/80)
[ ] prev/next file n p prev/next uncovered line
TLA Baseline Branch
Line Hits Code Line Hits Code
1   // 1   //
2   // Copyright (c) 2019 Vinnie Falco (vinnie.falco@gmail.com) 2   // Copyright (c) 2019 Vinnie Falco (vinnie.falco@gmail.com)
3   // Copyright (c) 2022 Alan de Freitas (alandefreitas@gmail.com) 3   // Copyright (c) 2022 Alan de Freitas (alandefreitas@gmail.com)
4   // 4   //
5   // Distributed under the Boost Software License, Version 1.0. (See accompanying 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) 6   // file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
7   // 7   //
8   // Official repository: https://github.com/boostorg/url 8   // Official repository: https://github.com/boostorg/url
9   // 9   //
10   10  
11   #ifndef BOOST_URL_IMPL_URL_BASE_HPP 11   #ifndef BOOST_URL_IMPL_URL_BASE_HPP
12   #define BOOST_URL_IMPL_URL_BASE_HPP 12   #define BOOST_URL_IMPL_URL_BASE_HPP
13   13  
14   #include <boost/url/encode.hpp> 14   #include <boost/url/encode.hpp>
15   #include <boost/url/error.hpp> 15   #include <boost/url/error.hpp>
16   #include <boost/url/host_type.hpp> 16   #include <boost/url/host_type.hpp>
17   #include <boost/url/scheme.hpp> 17   #include <boost/url/scheme.hpp>
18   #include <boost/url/url_view.hpp> 18   #include <boost/url/url_view.hpp>
19   #include <boost/url/detail/any_params_iter.hpp> 19   #include <boost/url/detail/any_params_iter.hpp>
20   #include <boost/url/detail/any_segments_iter.hpp> 20   #include <boost/url/detail/any_segments_iter.hpp>
21   #include <boost/url/detail/decode.hpp> 21   #include <boost/url/detail/decode.hpp>
22   #include <boost/url/detail/encode.hpp> 22   #include <boost/url/detail/encode.hpp>
23   #include <boost/url/detail/except.hpp> 23   #include <boost/url/detail/except.hpp>
24   #include <boost/url/detail/normalize.hpp> 24   #include <boost/url/detail/normalize.hpp>
25   #include <boost/url/detail/path.hpp> 25   #include <boost/url/detail/path.hpp>
26   #include <boost/url/detail/print.hpp> 26   #include <boost/url/detail/print.hpp>
27   #include <boost/url/grammar/ci_string.hpp> 27   #include <boost/url/grammar/ci_string.hpp>
28   #include <boost/url/rfc/authority_rule.hpp> 28   #include <boost/url/rfc/authority_rule.hpp>
29   #include <boost/url/rfc/query_rule.hpp> 29   #include <boost/url/rfc/query_rule.hpp>
30   #include <boost/url/rfc/ipv6_address_rule.hpp> 30   #include <boost/url/rfc/ipv6_address_rule.hpp>
31   #include <boost/url/rfc/detail/charsets.hpp> 31   #include <boost/url/rfc/detail/charsets.hpp>
32   #include <boost/url/rfc/detail/host_rule.hpp> 32   #include <boost/url/rfc/detail/host_rule.hpp>
33   #include <boost/url/rfc/detail/ipvfuture_rule.hpp> 33   #include <boost/url/rfc/detail/ipvfuture_rule.hpp>
34   #include <boost/url/rfc/detail/path_rules.hpp> 34   #include <boost/url/rfc/detail/path_rules.hpp>
35   #include <boost/url/rfc/detail/port_rule.hpp> 35   #include <boost/url/rfc/detail/port_rule.hpp>
36   #include <boost/url/rfc/detail/scheme_rule.hpp> 36   #include <boost/url/rfc/detail/scheme_rule.hpp>
37   #include <boost/url/rfc/detail/userinfo_rule.hpp> 37   #include <boost/url/rfc/detail/userinfo_rule.hpp>
38   #include <boost/url/grammar/parse.hpp> 38   #include <boost/url/grammar/parse.hpp>
39   #include <boost/url/detail/move_chars.hpp> 39   #include <boost/url/detail/move_chars.hpp>
40   #include <cstring> 40   #include <cstring>
41   #include <iostream> 41   #include <iostream>
42   #include <stdexcept> 42   #include <stdexcept>
43   #include <utility> 43   #include <utility>
44   44  
45   namespace boost { 45   namespace boost {
46   namespace urls { 46   namespace urls {
47   47  
48   //------------------------------------------------ 48   //------------------------------------------------
49   49  
50   // these objects help handle the cases 50   // these objects help handle the cases
51   // where the user passes in strings that 51   // where the user passes in strings that
52   // come from inside the url buffer. 52   // come from inside the url buffer.
53   53  
54   inline 54   inline
HITCBC 55   18347 url_base:: 55   18574 url_base::
56   op_t:: 56   op_t::
57   ~op_t() 57   ~op_t()
58   { 58   {
HITCBC 59   18347 if(old) 59   18574 if(old)
HITCBC 60   2567 u.cleanup(*this); 60   2575 u.cleanup(*this);
HITCBC 61   18347 u.check_invariants(); 61   18574 u.check_invariants();
HITCBC 62   18347 } 62   18574 }
63   63  
64   inline 64   inline
HITCBC 65   18347 url_base:: 65   18574 url_base::
66   op_t:: 66   op_t::
67   op_t( 67   op_t(
68   url_base& impl_, 68   url_base& impl_,
69   core::string_view* s0_, 69   core::string_view* s0_,
HITCBC 70   18347 core::string_view* s1_) noexcept 70   18574 core::string_view* s1_) noexcept
HITCBC 71   18347 : u(impl_) 71   18574 : u(impl_)
HITCBC 72   18347 , s0(s0_) 72   18574 , s0(s0_)
HITCBC 73   18347 , s1(s1_) 73   18574 , s1(s1_)
74   { 74   {
HITCBC 75   18347 u.check_invariants(); 75   18574 u.check_invariants();
HITCBC 76   18347 } 76   18574 }
77   77  
78   inline 78   inline
79   void 79   void
HITCBC 80   9410 url_base:: 80   9465 url_base::
81   op_t:: 81   op_t::
82   move( 82   move(
83   char* dest, 83   char* dest,
84   char const* src, 84   char const* src,
85   std::size_t n) noexcept 85   std::size_t n) noexcept
86   { 86   {
HITCBC 87   9410 if(! n) 87   9465 if(! n)
HITCBC 88   2523 return; 88   2523 return;
HITCBC 89   6887 if(s0) 89   6942 if(s0)
90   { 90   {
HITCBC 91   4905 if(s1) 91   4919 if(s1)
HITCBC 92   656 return detail::move_chars( 92   656 return detail::move_chars(
HITCBC 93   656 dest, src, n, *s0, *s1); 93   656 dest, src, n, *s0, *s1);
HITCBC 94   4249 return detail::move_chars( 94   4263 return detail::move_chars(
HITCBC 95   4249 dest, src, n, *s0); 95   4263 dest, src, n, *s0);
96   } 96   }
HITCBC 97   1982 detail::move_chars( 97   2023 detail::move_chars(
98   dest, src, n); 98   dest, src, n);
99   } 99   }
100   100  
101   //------------------------------------------------ 101   //------------------------------------------------
102   102  
103   // construct reference 103   // construct reference
104   inline 104   inline
HITCBC 105   1647 url_base:: 105   1647 url_base::
106   url_base( 106   url_base(
HITCBC 107   1647 detail::url_impl const& impl) noexcept 107   1647 detail::url_impl const& impl) noexcept
HITCBC 108   1647 : url_view_base(impl) 108   1647 : url_view_base(impl)
109   { 109   {
HITCBC 110   1647 } 110   1647 }
111   111  
112   inline 112   inline
113   void 113   void
HITCBC 114   308 url_base:: 114   308 url_base::
115   reserve_impl(std::size_t n) 115   reserve_impl(std::size_t n)
116   { 116   {
HITCBC 117   308 op_t op(*this); 117   308 op_t op(*this);
HITCBC 118   308 reserve_impl(n, op); 118   308 reserve_impl(n, op);
HITCBC 119   307 if(s_) 119   307 if(s_)
HITCBC 120   305 s_[size()] = '\0'; 120   305 s_[size()] = '\0';
HITCBC 121   308 } 121   308 }
122   122  
123   // make a copy of u 123   // make a copy of u
124   inline 124   inline
125   void 125   void
HITCBC 126   5241 url_base:: 126   5295 url_base::
127   copy(url_view_base const& u) 127   copy(url_view_base const& u)
128   { 128   {
HITCBC 129   5241 if (this == &u) 129   5295 if (this == &u)
HITCBC 130   140 return; 130   140 return;
HITCBC 131   5227 op_t op(*this); 131   5281 op_t op(*this);
HITCBC 132   5227 if(u.size() == 0) 132   5281 if(u.size() == 0)
133   { 133   {
HITCBC 134   126 clear(); 134   126 clear();
HITCBC 135   126 return; 135   126 return;
136   } 136   }
HITCBC 137   5101 reserve_impl( 137   5155 reserve_impl(
138   u.size(), op); 138   u.size(), op);
HITCBC 139   5098 impl_ = u.impl(); 139   5152 impl_ = u.impl();
HITCBC 140   5098 impl_.cs_ = s_; 140   5152 impl_.cs_ = s_;
HITCBC 141   5098 impl_.from_ = {from::url}; 141   5152 impl_.from_ = {from::url};
HITCBC 142   5098 std::memcpy(s_, 142   5152 std::memcpy(s_,
HITCBC 143   5098 u.data(), u.size()); 143   5152 u.data(), u.size());
HITCBC 144   5098 s_[size()] = '\0'; 144   5152 s_[size()] = '\0';
HITCBC 145   5227 } 145   5281 }
146   146  
147   //------------------------------------------------ 147   //------------------------------------------------
148   // 148   //
149   // Scheme 149   // Scheme
150   // 150   //
151   //------------------------------------------------ 151   //------------------------------------------------
152   152  
153   inline 153   inline
154   url_base& 154   url_base&
HITCBC 155   376 url_base:: 155   379 url_base::
156   set_scheme(core::string_view s) 156   set_scheme(core::string_view s)
157   { 157   {
HITCBC 158   376 set_scheme_impl( 158   379 set_scheme_impl(
159   s, string_to_scheme(s)); 159   s, string_to_scheme(s));
HITCBC 160   359 return *this; 160   362 return *this;
161   } 161   }
162   162  
163   inline 163   inline
164   url_base& 164   url_base&
HITCBC 165   182 url_base:: 165   182 url_base::
166   set_scheme_id(urls::scheme id) 166   set_scheme_id(urls::scheme id)
167   { 167   {
HITCBC 168   182 if(id == urls::scheme::unknown) 168   182 if(id == urls::scheme::unknown)
HITCBC 169   14 detail::throw_invalid_argument(); 169   14 detail::throw_invalid_argument();
HITCBC 170   168 if(id == urls::scheme::none) 170   168 if(id == urls::scheme::none)
HITCBC 171   18 return remove_scheme(); 171   18 return remove_scheme();
HITCBC 172   150 set_scheme_impl(to_string(id), id); 172   150 set_scheme_impl(to_string(id), id);
HITCBC 173   119 return *this; 173   119 return *this;
174   } 174   }
175   175  
176   inline 176   inline
177   url_base& 177   url_base&
HITCBC 178   53 url_base:: 178   56 url_base::
179   remove_scheme() 179   remove_scheme()
180   { 180   {
HITCBC 181   53 op_t op(*this); 181   56 op_t op(*this);
HITCBC 182   53 auto const sn = impl_.len(id_scheme); 182   56 auto const sn = impl_.len(id_scheme);
HITCBC 183   53 if(sn == 0) 183   56 if(sn == 0)
HITCBC 184   15 return *this; 184   15 return *this;
HITCBC 185   38 auto const po = impl_.offset(id_path); 185   41 auto const po = impl_.offset(id_path);
ECB 186 - 38 auto fseg = first_segment(); 186 + // The first segment as written, not as
  187 + // first_segment() returns it: a "./" prefix
  188 + // already keeps a colon in the next segment
  189 + // from looking like a scheme, and the colon
  190 + // encoding below starts at the path
HITGNC   191 + 41 core::string_view fseg = impl_.get(id_path);
HITGNC   192 + 41 fseg = fseg.substr(0, fseg.find('/'));
187   bool const encode_colon = 193   bool const encode_colon =
HITCBC 188   38 !has_authority() && 194   41 !has_authority() &&
HITCBC 189   21 impl_.nseg_ > 0 && 195   24 impl_.nseg_ > 0 &&
HITCBC 190   70 s_[po] != '/' && 196   79 s_[po] != '/' &&
HITCBC 191   11 fseg.contains(':'); 197   14 fseg.contains(':');
HITCBC 192   38 if(!encode_colon) 198   41 if(!encode_colon)
193   { 199   {
194   // just remove the scheme 200   // just remove the scheme
HITCBC 195   29 resize_impl(id_scheme, 0, op); 201   32 resize_impl(id_scheme, 0, op);
HITCBC 196   29 impl_.scheme_ = urls::scheme::none; 202   32 impl_.scheme_ = urls::scheme::none;
HITCBC 197   29 check_invariants(); 203   32 check_invariants();
HITCBC 198   29 return *this; 204   32 return *this;
199   } 205   }
200   // encode any ":" in the first path segment 206   // encode any ":" in the first path segment
HITCBC 201   9 BOOST_ASSERT(sn >= 2); 207   9 BOOST_ASSERT(sn >= 2);
HITCBC 202   9 auto pn = impl_.len(id_path); 208   9 auto pn = impl_.len(id_path);
HITCBC 203   9 std::size_t cn = 0; 209   9 std::size_t cn = 0;
HITCBC 204   46 for (char c: fseg) 210   46 for (char c: fseg)
HITCBC 205   37 cn += c == ':'; 211   37 cn += c == ':';
206   std::size_t new_size = 212   std::size_t new_size =
HITCBC 207   9 size() - sn + 2 * cn; 213   9 size() - sn + 2 * cn;
HITCBC 208   9 bool need_resize = new_size > size(); 214   9 bool need_resize = new_size > size();
HITCBC 209   9 if (need_resize) 215   9 if (need_resize)
210   { 216   {
HITCBC 211   1 resize_impl( 217   1 resize_impl(
HITCBC 212   1 id_path, pn + 2 * cn, op); 218   1 id_path, pn + 2 * cn, op);
213   } 219   }
214   // move [id_scheme, id_path) left 220   // move [id_scheme, id_path) left
HITCBC 215   9 op.move( 221   9 op.move(
216   s_, 222   s_,
HITCBC 217   9 s_ + sn, 223   9 s_ + sn,
218   po - sn); 224   po - sn);
219   // move [id_path, id_query) left 225   // move [id_path, id_query) left
HITCBC 220   9 auto qo = impl_.offset(id_query); 226   9 auto qo = impl_.offset(id_query);
HITCBC 221   9 op.move( 227   9 op.move(
HITCBC 222   9 s_ + po - sn, 228   9 s_ + po - sn,
HITCBC 223   9 s_ + po, 229   9 s_ + po,
224   qo - po); 230   qo - po);
225   // move [id_query, id_end) left 231   // move [id_query, id_end) left
HITCBC 226   9 op.move( 232   9 op.move(
HITCBC 227   9 s_ + qo - sn + 2 * cn, 233   9 s_ + qo - sn + 2 * cn,
HITCBC 228   9 s_ + qo, 234   9 s_ + qo,
HITCBC 229   9 impl_.offset(id_end) - qo); 235   9 impl_.offset(id_end) - qo);
230   236  
231   // adjust part offsets. 237   // adjust part offsets.
232   // (po and qo are invalidated) 238   // (po and qo are invalidated)
HITCBC 233   9 if (need_resize) 239   9 if (need_resize)
234   { 240   {
HITCBC 235   1 impl_.adjust_left(id_user, id_end, sn); 241   1 impl_.adjust_left(id_user, id_end, sn);
236   } 242   }
237   else 243   else
238   { 244   {
HITCBC 239   8 impl_.adjust_left(id_user, id_path, sn); 245   8 impl_.adjust_left(id_user, id_path, sn);
HITCBC 240   8 impl_.adjust_left(id_query, id_end, sn - 2 * cn); 246   8 impl_.adjust_left(id_query, id_end, sn - 2 * cn);
241   } 247   }
HITCBC 242   9 if (encode_colon) 248   9 if (encode_colon)
243   { 249   {
244   // move the 2nd, 3rd, ... segments 250   // move the 2nd, 3rd, ... segments
HITCBC 245   9 auto begin = s_ + impl_.offset(id_path); 251   9 auto begin = s_ + impl_.offset(id_path);
HITCBC 246   9 auto it = begin; 252   9 auto it = begin;
HITCBC 247   9 auto end = begin + pn; 253   9 auto end = begin + pn;
HITCBC 248   46 while (it != end && 254   46 while (it != end &&
HITCBC 249   40 *it != '/') 255   40 *it != '/')
HITCBC 250   37 ++it; 256   37 ++it;
251   // we don't need op here because this is 257   // we don't need op here because this is
252   // an internal operation 258   // an internal operation
HITCBC 253   9 std::memmove(it + (2 * cn), it, end - it); 259   9 std::memmove(it + (2 * cn), it, end - it);
254   260  
255   // move 1st segment 261   // move 1st segment
HITCBC 256   9 auto src = s_ + impl_.offset(id_path) + pn; 262   9 auto src = s_ + impl_.offset(id_path) + pn;
HITCBC 257   9 auto dest = s_ + impl_.offset(id_query); 263   9 auto dest = s_ + impl_.offset(id_query);
HITCBC 258   9 src -= end - it; 264   9 src -= end - it;
HITCBC 259   9 dest -= end - it; 265   9 dest -= end - it;
HITCBC 260   9 pn -= end - it; 266   9 pn -= end - it;
261   do { 267   do {
HITCBC 262   37 --src; 268   37 --src;
HITCBC 263   37 --dest; 269   37 --dest;
HITCBC 264   37 if (*src != ':') 270   37 if (*src != ':')
265   { 271   {
HITCBC 266   25 *dest = *src; 272   25 *dest = *src;
267   } 273   }
268   else 274   else
269   { 275   {
270   // use uppercase as required by 276   // use uppercase as required by
271   // syntax-based normalization 277   // syntax-based normalization
HITCBC 272   12 *dest-- = 'A'; 278   12 *dest-- = 'A';
HITCBC 273   12 *dest-- = '3'; 279   12 *dest-- = '3';
HITCBC 274   12 *dest = '%'; 280   12 *dest = '%';
275   } 281   }
HITCBC 276   37 --pn; 282   37 --pn;
HITCBC 277   37 } while (pn); 283   37 } while (pn);
278   } 284   }
HITCBC 279   9 s_[size()] = '\0'; 285   9 s_[size()] = '\0';
HITCBC 280   9 impl_.scheme_ = urls::scheme::none; 286   9 impl_.scheme_ = urls::scheme::none;
HITCBC 281   9 return *this; 287   9 return *this;
HITCBC 282   53 } 288   56 }
283   289  
284   //------------------------------------------------ 290   //------------------------------------------------
285   // 291   //
286   // Authority 292   // Authority
287   // 293   //
288   //------------------------------------------------ 294   //------------------------------------------------
289   295  
290   inline 296   inline
291   url_base& 297   url_base&
HITCBC 292   120 url_base:: 298   124 url_base::
293   set_encoded_authority( 299   set_encoded_authority(
294   pct_string_view s) 300   pct_string_view s)
295   { 301   {
HITCBC 296   120 op_t op(*this, &detail::ref(s)); 302   124 op_t op(*this, &detail::ref(s));
HITCBC 297   122 authority_view a = grammar::parse( 303   126 authority_view a = grammar::parse(
298   s, authority_rule 304   s, authority_rule
HITCBC 299   120 ).value(BOOST_URL_POS); 305   124 ).value(BOOST_URL_POS);
HITCBC 300   119 auto n = s.size() + 2; 306   123 auto n = s.size() + 2;
301   auto const need_slash = 307   auto const need_slash =
HITCBC 302   141 ! is_path_absolute() && 308   149 ! is_path_absolute() &&
HITCBC 303   22 impl_.len(id_path) > 0; 309   26 impl_.len(id_path) > 0;
HITCBC 304   119 if(need_slash) 310   123 if(need_slash)
HITCBC 305   2 ++n; 311   6 ++n;
HITCBC 306   119 auto dest = resize_impl( 312   123 auto dest = resize_impl(
307   id_user, id_path, n, op); 313   id_user, id_path, n, op);
HITCBC 308   119 dest[0] = '/'; 314   123 dest[0] = '/';
HITCBC 309   119 dest[1] = '/'; 315   123 dest[1] = '/';
HITCBC 310   119 std::memcpy(dest + 2, 316   123 std::memcpy(dest + 2,
HITCBC 311   119 s.data(), s.size()); 317   123 s.data(), s.size());
HITCBC 312   119 if(need_slash) 318   123 if(need_slash)
HITCBC 313   2 dest[n - 1] = '/'; 319   6 dest[n - 1] = '/';
HITCBC 314   119 impl_.apply_authority(a.u_); 320   123 impl_.apply_authority(a.u_);
HITCBC 315   119 if(need_slash) 321   123 if(need_slash)
  322 + {
HITCBC 316   2 impl_.adjust_right( 323   6 impl_.adjust_right(
317   id_query, id_end, 1); 324   id_query, id_end, 1);
HITGNC   325 + 6 ++impl_.decoded_[id_path];
  326 + }
HITCBC 318   119 return *this; 327   123 return *this;
HITCBC 319   120 } 328   124 }
320   329  
321   inline 330   inline
322   url_base& 331   url_base&
HITCBC 323   260 url_base:: 332   263 url_base::
324   remove_authority() 333   remove_authority()
325   { 334   {
HITCBC 326   260 if(! has_authority()) 335   263 if(! has_authority())
HITCBC 327   70 return *this; 336   70 return *this;
328   337  
HITCBC 329   190 op_t op(*this); 338   193 op_t op(*this);
HITCBC 330   190 auto path = impl_.get(id_path); 339   193 auto path = impl_.get(id_path);
HITCBC 331   190 bool const need_dot = path.starts_with("//"); 340   193 bool const need_dot = path.starts_with("//");
HITCBC 332   190 if(need_dot) 341   193 if(need_dot)
333   { 342   {
334   // prepend "/.", can't throw 343   // prepend "/.", can't throw
HITCBC 335   5 auto p = resize_impl( 344   8 auto p = resize_impl(
336   id_user, id_path, 2, op); 345   id_user, id_path, 2, op);
HITCBC 337   5 p[0] = '/'; 346   8 p[0] = '/';
HITCBC 338   5 p[1] = '.'; 347   8 p[1] = '.';
HITCBC 339   5 impl_.split(id_user, 0); 348   8 impl_.split(id_user, 0);
HITCBC 340   5 impl_.split(id_pass, 0); 349   8 impl_.split(id_pass, 0);
HITCBC 341   5 impl_.split(id_host, 0); 350   8 impl_.split(id_host, 0);
HITCBC 342   5 impl_.split(id_port, 0); 351   8 impl_.split(id_port, 0);
HITGNC   352 + 8 impl_.decoded_[id_path] += 2;
343   } 353   }
344   else 354   else
345   { 355   {
HITCBC 346   185 resize_impl( 356   185 resize_impl(
347   id_user, id_path, 0, op); 357   id_user, id_path, 0, op);
348   } 358   }
HITCBC 349   190 impl_.decoded_[id_user] = 0; 359   193 impl_.decoded_[id_user] = 0;
HITCBC 350   190 impl_.decoded_[id_pass] = 0; 360   193 impl_.decoded_[id_pass] = 0;
HITCBC 351   190 impl_.decoded_[id_host] = 0; 361   193 impl_.decoded_[id_host] = 0;
HITCBC 352   190 impl_.decoded_[id_port] = 0; 362   193 impl_.decoded_[id_port] = 0;
HITCBC 353   190 impl_.host_type_ = 363   193 impl_.host_type_ =
354   urls::host_type::none; 364   urls::host_type::none;
HITCBC 355   190 return *this; 365   193 return *this;
HITCBC 356   190 } 366   193 }
357   367  
358   //------------------------------------------------ 368   //------------------------------------------------
359   // 369   //
360   // Userinfo 370   // Userinfo
361   // 371   //
362   //------------------------------------------------ 372   //------------------------------------------------
363   373  
364   inline 374   inline
365   url_base& 375   url_base&
HITCBC 366   225 url_base:: 376   225 url_base::
367   set_userinfo( 377   set_userinfo(
368   core::string_view s) 378   core::string_view s)
369   { 379   {
HITCBC 370   225 op_t op(*this, &s); 380   225 op_t op(*this, &s);
HITCBC 371   225 encoding_opts opt; 381   225 encoding_opts opt;
HITCBC 372   225 auto const n = encoded_size( 382   225 auto const n = encoded_size(
373   s, detail::userinfo_chars, opt); 383   s, detail::userinfo_chars, opt);
HITCBC 374   225 auto dest = set_userinfo_impl(n, op); 384   225 auto dest = set_userinfo_impl(n, op);
HITCBC 375   225 encode( 385   225 encode(
376   dest, 386   dest,
377   n, 387   n,
378   s, 388   s,
379   detail::userinfo_chars, 389   detail::userinfo_chars,
380   opt); 390   opt);
HITCBC 381   225 auto const pos = impl_.get( 391   225 auto const pos = impl_.get(
382   id_user, id_host 392   id_user, id_host
HITCBC 383   225 ).find_first_of(':'); 393   225 ).find_first_of(':');
HITCBC 384   225 if(pos != core::string_view::npos) 394   225 if(pos != core::string_view::npos)
385   { 395   {
HITCBC 386   22 impl_.split(id_user, pos); 396   22 impl_.split(id_user, pos);
387   // find ':' in plain string 397   // find ':' in plain string
388   auto const pos2 = 398   auto const pos2 =
HITCBC 389   22 s.find_first_of(':'); 399   22 s.find_first_of(':');
HITCBC 390   22 if(pos2 != core::string_view::npos) 400   22 if(pos2 != core::string_view::npos)
391   { 401   {
392   // pos2 is the ':' index in plain input (user[:pass]) 402   // pos2 is the ':' index in plain input (user[:pass])
393   // decoded user is [0, pos2), decoded pass is (pos2, end]. 403   // decoded user is [0, pos2), decoded pass is (pos2, end].
HITCBC 394   22 impl_.decoded_[id_user] = 404   22 impl_.decoded_[id_user] =
HITCBC 395   22 detail::to_size_type(pos2); 405   22 detail::to_size_type(pos2);
HITCBC 396   22 impl_.decoded_[id_pass] = 406   22 impl_.decoded_[id_pass] =
HITCBC 397   22 detail::to_size_type(s.size() - pos2 - 1); 407   22 detail::to_size_type(s.size() - pos2 - 1);
398   } 408   }
399   else 409   else
400   { 410   {
MISUBC 401   ✗ impl_.decoded_[id_user] = 411   ✗ impl_.decoded_[id_user] =
MISUBC 402   ✗ detail::to_size_type(s.size()); 412   ✗ detail::to_size_type(s.size());
MISUBC 403   ✗ impl_.decoded_[id_pass] = 0; 413   ✗ impl_.decoded_[id_pass] = 0;
404   } 414   }
405   } 415   }
406   else 416   else
407   { 417   {
HITCBC 408   203 impl_.decoded_[id_user] = 418   203 impl_.decoded_[id_user] =
HITCBC 409   203 detail::to_size_type(s.size()); 419   203 detail::to_size_type(s.size());
HITCBC 410   203 impl_.decoded_[id_pass] = 0; 420   203 impl_.decoded_[id_pass] = 0;
411   } 421   }
HITCBC 412   225 return *this; 422   225 return *this;
HITCBC 413   225 } 423   225 }
414   424  
415   inline 425   inline
416   url_base& 426   url_base&
HITCBC 417   52 url_base:: 427   53 url_base::
418   set_encoded_userinfo( 428   set_encoded_userinfo(
419   pct_string_view s) 429   pct_string_view s)
420   { 430   {
HITCBC 421   52 op_t op(*this, &detail::ref(s)); 431   53 op_t op(*this, &detail::ref(s));
HITCBC 422   52 auto const pos = s.find_first_of(':'); 432   53 auto const pos = s.find_first_of(':');
HITCBC 423   52 if(pos != core::string_view::npos) 433   53 if(pos != core::string_view::npos)
424   { 434   {
425   // user:pass 435   // user:pass
HITCBC 426   7 auto const s0 = s.substr(0, pos); 436   8 auto const s0 = s.substr(0, pos);
HITCBC 427   7 auto const s1 = s.substr(pos + 1); 437   8 auto const s1 = s.substr(pos + 1);
428   auto const n0 = 438   auto const n0 =
HITCBC 429   7 detail::re_encoded_size_unsafe( 439   8 detail::re_encoded_size_unsafe(
430   s0, 440   s0,
431   detail::user_chars); 441   detail::user_chars);
432   auto const n1 = 442   auto const n1 =
HITCBC 433   7 detail::re_encoded_size_unsafe(s1, 443   8 detail::re_encoded_size_unsafe(s1,
434   detail::password_chars); 444   detail::password_chars);
435   auto dest = 445   auto dest =
HITCBC 436   7 set_userinfo_impl(n0 + n1 + 1, op); 446   8 set_userinfo_impl(n0 + n1 + 1, op);
HITCBC 437   7 impl_.decoded_[id_user] = 447   8 impl_.decoded_[id_user] =
HITCBC 438   7 detail::to_size_type(detail::re_encode_unsafe( 448   8 detail::to_size_type(detail::re_encode_unsafe(
439   dest, 449   dest,
HITCBC 440   7 dest + n0, 450   8 dest + n0,
441   s0, 451   s0,
442   detail::user_chars)); 452   detail::user_chars));
HITCBC 443   7 *dest++ = ':'; 453   8 *dest++ = ':';
HITCBC 444   7 impl_.decoded_[id_pass] = 454   8 impl_.decoded_[id_pass] =
HITCBC 445   7 detail::to_size_type(detail::re_encode_unsafe( 455   8 detail::to_size_type(detail::re_encode_unsafe(
446   dest, 456   dest,
HITCBC 447   7 dest + n1, 457   8 dest + n1,
448   s1, 458   s1,
449   detail::password_chars)); 459   detail::password_chars));
HITCBC 450   7 impl_.split(id_user, 2 + n0); 460   8 impl_.split(id_user, 2 + n0);
451   } 461   }
452   else 462   else
453   { 463   {
454   // user 464   // user
455   auto const n = 465   auto const n =
HITCBC 456   45 detail::re_encoded_size_unsafe( 466   45 detail::re_encoded_size_unsafe(
457   s, detail::user_chars); 467   s, detail::user_chars);
HITCBC 458   45 auto dest = set_userinfo_impl(n, op); 468   45 auto dest = set_userinfo_impl(n, op);
HITCBC 459   45 impl_.decoded_[id_user] = 469   45 impl_.decoded_[id_user] =
HITCBC 460   90 detail::to_size_type(detail::re_encode_unsafe( 470   90 detail::to_size_type(detail::re_encode_unsafe(
461   dest, 471   dest,
HITCBC 462   45 dest + n, 472   45 dest + n,
463   s, 473   s,
464   detail::user_chars)); 474   detail::user_chars));
HITCBC 465   45 impl_.split(id_user, 2 + n); 475   45 impl_.split(id_user, 2 + n);
HITCBC 466   45 impl_.decoded_[id_pass] = 0; 476   45 impl_.decoded_[id_pass] = 0;
467   } 477   }
HITCBC 468   52 return *this; 478   53 return *this;
HITCBC 469   52 } 479   53 }
470   480  
471   inline 481   inline
472   url_base& 482   url_base&
HITCBC 473   214 url_base:: 483   214 url_base::
474   remove_userinfo() noexcept 484   remove_userinfo() noexcept
475   { 485   {
HITCBC 476   214 if(impl_.len(id_pass) == 0) 486   214 if(impl_.len(id_pass) == 0)
HITCBC 477   138 return *this; // no userinfo 487   138 return *this; // no userinfo
478   488  
HITCBC 479   76 op_t op(*this); 489   76 op_t op(*this);
480   // keep authority '//' 490   // keep authority '//'
HITCBC 481   76 resize_impl( 491   76 resize_impl(
482   id_user, id_host, 2, op); 492   id_user, id_host, 2, op);
HITCBC 483   76 impl_.decoded_[id_user] = 0; 493   76 impl_.decoded_[id_user] = 0;
HITCBC 484   76 impl_.decoded_[id_pass] = 0; 494   76 impl_.decoded_[id_pass] = 0;
HITCBC 485   76 return *this; 495   76 return *this;
HITCBC 486   76 } 496   76 }
487   497  
488   //------------------------------------------------ 498   //------------------------------------------------
489   499  
490   inline 500   inline
491   url_base& 501   url_base&
HITCBC 492   241 url_base:: 502   241 url_base::
493   set_user(core::string_view s) 503   set_user(core::string_view s)
494   { 504   {
HITCBC 495   241 op_t op(*this, &s); 505   241 op_t op(*this, &s);
HITCBC 496   241 encoding_opts opt; 506   241 encoding_opts opt;
HITCBC 497   241 auto const n = encoded_size( 507   241 auto const n = encoded_size(
498   s, detail::user_chars, opt); 508   s, detail::user_chars, opt);
HITCBC 499   241 auto dest = set_user_impl(n, op); 509   241 auto dest = set_user_impl(n, op);
HITCBC 500   241 encode_unsafe( 510   241 encode_unsafe(
501   dest, 511   dest,
502   n, 512   n,
503   s, 513   s,
504   detail::user_chars, 514   detail::user_chars,
505   opt); 515   opt);
HITCBC 506   241 impl_.decoded_[id_user] = 516   241 impl_.decoded_[id_user] =
HITCBC 507   241 detail::to_size_type(s.size()); 517   241 detail::to_size_type(s.size());
HITCBC 508   241 return *this; 518   241 return *this;
HITCBC 509   241 } 519   241 }
510   520  
511   inline 521   inline
512   url_base& 522   url_base&
HITCBC 513   147 url_base:: 523   148 url_base::
514   set_encoded_user( 524   set_encoded_user(
515   pct_string_view s) 525   pct_string_view s)
516   { 526   {
HITCBC 517   147 op_t op(*this, &detail::ref(s)); 527   148 op_t op(*this, &detail::ref(s));
518   auto const n = 528   auto const n =
HITCBC 519   147 detail::re_encoded_size_unsafe( 529   148 detail::re_encoded_size_unsafe(
520   s, detail::user_chars); 530   s, detail::user_chars);
HITCBC 521   147 auto dest = set_user_impl(n, op); 531   148 auto dest = set_user_impl(n, op);
HITCBC 522   147 impl_.decoded_[id_user] = 532   148 impl_.decoded_[id_user] =
HITCBC 523   294 detail::to_size_type(detail::re_encode_unsafe( 533   296 detail::to_size_type(detail::re_encode_unsafe(
524   dest, 534   dest,
HITCBC 525   147 dest + n, 535   148 dest + n,
526   s, 536   s,
527   detail::user_chars)); 537   detail::user_chars));
HITCBC 528   147 BOOST_ASSERT( 538   148 BOOST_ASSERT(
529   impl_.decoded_[id_user] == 539   impl_.decoded_[id_user] ==
530   s.decoded_size()); 540   s.decoded_size());
HITCBC 531   147 return *this; 541   148 return *this;
HITCBC 532   147 } 542   148 }
533   543  
534   //------------------------------------------------ 544   //------------------------------------------------
535   545  
536   inline 546   inline
537   url_base& 547   url_base&
HITCBC 538   231 url_base:: 548   231 url_base::
539   set_password(core::string_view s) 549   set_password(core::string_view s)
540   { 550   {
HITCBC 541   231 op_t op(*this, &s); 551   231 op_t op(*this, &s);
HITCBC 542   231 encoding_opts opt; 552   231 encoding_opts opt;
HITCBC 543   231 auto const n = encoded_size( 553   231 auto const n = encoded_size(
544   s, detail::password_chars, opt); 554   s, detail::password_chars, opt);
HITCBC 545   231 auto dest = set_password_impl(n, op); 555   231 auto dest = set_password_impl(n, op);
HITCBC 546   231 encode_unsafe( 556   231 encode_unsafe(
547   dest, 557   dest,
548   n, 558   n,
549   s, 559   s,
550   detail::password_chars, 560   detail::password_chars,
551   opt); 561   opt);
HITCBC 552   231 impl_.decoded_[id_pass] = 562   231 impl_.decoded_[id_pass] =
HITCBC 553   231 detail::to_size_type(s.size()); 563   231 detail::to_size_type(s.size());
HITCBC 554   231 return *this; 564   231 return *this;
HITCBC 555   231 } 565   231 }
556   566  
557   inline 567   inline
558   url_base& 568   url_base&
HITCBC 559   140 url_base:: 569   141 url_base::
560   set_encoded_password( 570   set_encoded_password(
561   pct_string_view s) 571   pct_string_view s)
562   { 572   {
HITCBC 563   140 op_t op(*this, &detail::ref(s)); 573   141 op_t op(*this, &detail::ref(s));
564   auto const n = 574   auto const n =
HITCBC 565   140 detail::re_encoded_size_unsafe( 575   141 detail::re_encoded_size_unsafe(
566   s, 576   s,
567   detail::password_chars); 577   detail::password_chars);
HITCBC 568   140 auto dest = set_password_impl(n, op); 578   141 auto dest = set_password_impl(n, op);
HITCBC 569   140 impl_.decoded_[id_pass] = 579   141 impl_.decoded_[id_pass] =
HITCBC 570   280 detail::to_size_type(detail::re_encode_unsafe( 580   282 detail::to_size_type(detail::re_encode_unsafe(
571   dest, 581   dest,
HITCBC 572   140 dest + n, 582   141 dest + n,
573   s, 583   s,
574   detail::password_chars)); 584   detail::password_chars));
HITCBC 575   140 BOOST_ASSERT( 585   141 BOOST_ASSERT(
576   impl_.decoded_[id_pass] == 586   impl_.decoded_[id_pass] ==
577   s.decoded_size()); 587   s.decoded_size());
HITCBC 578   140 return *this; 588   141 return *this;
HITCBC 579   140 } 589   141 }
580   590  
581   inline 591   inline
582   url_base& 592   url_base&
HITCBC 583   19 url_base:: 593   19 url_base::
584   remove_password() noexcept 594   remove_password() noexcept
585   { 595   {
HITCBC 586   19 auto const n = impl_.len(id_pass); 596   19 auto const n = impl_.len(id_pass);
HITCBC 587   19 if(n < 2) 597   19 if(n < 2)
HITCBC 588   12 return *this; // no password 598   12 return *this; // no password
589   599  
HITCBC 590   7 op_t op(*this); 600   7 op_t op(*this);
591   // clear password, retain '@' 601   // clear password, retain '@'
592   auto dest = 602   auto dest =
HITCBC 593   7 resize_impl(id_pass, 1, op); 603   7 resize_impl(id_pass, 1, op);
HITCBC 594   7 dest[0] = '@'; 604   7 dest[0] = '@';
HITCBC 595   7 impl_.decoded_[id_pass] = 0; 605   7 impl_.decoded_[id_pass] = 0;
HITCBC 596   7 return *this; 606   7 return *this;
HITCBC 597   7 } 607   7 }
598   608  
599   //------------------------------------------------ 609   //------------------------------------------------
600   // 610   //
601   // Host 611   // Host
602   // 612   //
603   //------------------------------------------------ 613   //------------------------------------------------
604   /* 614   /*
605   host_type host_type() // ipv4, ipv6, ipvfuture, name 615   host_type host_type() // ipv4, ipv6, ipvfuture, name
606   616  
607   std::string host() // return encoded_host().decode() 617   std::string host() // return encoded_host().decode()
608   pct_string_view encoded_host() // return host part, as-is 618   pct_string_view encoded_host() // return host part, as-is
609   std::string host_address() // return encoded_host_address().decode() 619   std::string host_address() // return encoded_host_address().decode()
610   pct_string_view encoded_host_address() // ipv4, ipv6, ipvfut, or encoded name, no brackets 620   pct_string_view encoded_host_address() // ipv4, ipv6, ipvfut, or encoded name, no brackets
611   621  
612   ipv4_address host_ipv4_address() // return ipv4_address or {} 622   ipv4_address host_ipv4_address() // return ipv4_address or {}
613   ipv6_address host_ipv6_address() // return ipv6_address or {} 623   ipv6_address host_ipv6_address() // return ipv6_address or {}
614   core::string_view host_ipvfuture() // return ipvfuture or {} 624   core::string_view host_ipvfuture() // return ipvfuture or {}
615   std::string host_name() // return decoded name or "" 625   std::string host_name() // return decoded name or ""
616   pct_string_view encoded_host_name() // return encoded host name or "" 626   pct_string_view encoded_host_name() // return encoded host name or ""
617   627  
618   -------------------------------------------------- 628   --------------------------------------------------
619   629  
620   set_host( core::string_view ) // set host part from plain text 630   set_host( core::string_view ) // set host part from plain text
621   set_encoded_host( pct_string_view ) // set host part from encoded text 631   set_encoded_host( pct_string_view ) // set host part from encoded text
622   set_host_address( core::string_view ) // set host from ipv4, ipv6, ipvfut, or plain reg-name string 632   set_host_address( core::string_view ) // set host from ipv4, ipv6, ipvfut, or plain reg-name string
623   set_encoded_host_address( pct_string_view ) // set host from ipv4, ipv6, ipvfut, or encoded reg-name string 633   set_encoded_host_address( pct_string_view ) // set host from ipv4, ipv6, ipvfut, or encoded reg-name string
624   634  
625   set_host_ipv4( ipv4_address ) // set ipv4 635   set_host_ipv4( ipv4_address ) // set ipv4
626   set_host_ipv6( ipv6_address ) // set ipv6 636   set_host_ipv6( ipv6_address ) // set ipv6
627   set_host_ipvfuture( core::string_view ) // set ipvfuture 637   set_host_ipvfuture( core::string_view ) // set ipvfuture
628   set_host_name( core::string_view ) // set name from plain 638   set_host_name( core::string_view ) // set name from plain
629   set_encoded_host_name( pct_string_view ) // set name from encoded 639   set_encoded_host_name( pct_string_view ) // set name from encoded
630   */ 640   */
631   641  
632   // set host part from plain text 642   // set host part from plain text
633   inline 643   inline
634   url_base& 644   url_base&
HITCBC 635   404 url_base:: 645   404 url_base::
636   set_host( 646   set_host(
637   core::string_view s) 647   core::string_view s)
638   { 648   {
HITCBC 639   404 if( s.size() > 2 && 649   404 if( s.size() > 2 &&
HITCBC 640   456 s.front() == '[' && 650   456 s.front() == '[' &&
HITCBC 641   52 s.back() == ']') 651   52 s.back() == ']')
642   { 652   {
643   // IP-literal 653   // IP-literal
HITCBC 644   52 if (s[1] != 'v') 654   52 if (s[1] != 'v')
645   { 655   {
646   // IPv6-address 656   // IPv6-address
HITCBC 647   51 auto innersv = s.substr(1, s.size() - 2); 657   51 auto innersv = s.substr(1, s.size() - 2);
HITCBC 648   51 auto innerit = innersv.begin(); 658   51 auto innerit = innersv.begin();
HITCBC 649   51 auto endit = innersv.end(); 659   51 auto endit = innersv.end();
HITCBC 650   51 auto rv = grammar::parse( 660   51 auto rv = grammar::parse(
651   innerit, 661   innerit,
652   endit, 662   endit,
653   ipv6_address_rule); 663   ipv6_address_rule);
HITCBC 654   51 if(rv) 664   51 if(rv)
655   { 665   {
HITCBC 656   51 if (innerit == endit) 666   51 if (innerit == endit)
657   { 667   {
HITCBC 658   47 set_host_ipv6_and_encoded_zone_id(*rv, {}); 668   47 set_host_ipv6_and_encoded_zone_id(*rv, {});
HITCBC 659   51 return *this; 669   51 return *this;
660   } 670   }
661   // IPv6addrz: https://datatracker.ietf.org/doc/html/rfc6874 671   // IPv6addrz: https://datatracker.ietf.org/doc/html/rfc6874
HITCBC 662   4 auto chars_left = endit - innerit; 672   4 auto chars_left = endit - innerit;
HITCBC 663   8 if (chars_left >= 2 && 673   8 if (chars_left >= 2 &&
HITCBC 664   4 *innerit++ == '%') 674   4 *innerit++ == '%')
665   { 675   {
HITCBC 666   4 core::string_view zone_id_str = {&*innerit, std::size_t(chars_left - 1)}; 676   4 core::string_view zone_id_str = {&*innerit, std::size_t(chars_left - 1)};
HITCBC 667   4 set_host_ipv6_and_zone_id(*rv, zone_id_str); 677   4 set_host_ipv6_and_zone_id(*rv, zone_id_str);
HITCBC 668   4 return *this; 678   4 return *this;
669   } 679   }
670   } 680   }
671   } 681   }
672   else 682   else
673   { 683   {
674   // IPvFuture 684   // IPvFuture
HITCBC 675   1 auto rv = grammar::parse( 685   1 auto rv = grammar::parse(
HITCBC 676   1 s.substr(1, s.size() - 2), 686   1 s.substr(1, s.size() - 2),
677   detail::ipvfuture_rule); 687   detail::ipvfuture_rule);
HITCBC 678   1 if(rv) 688   1 if(rv)
HITCBC 679   1 return set_host_ipvfuture(rv->str); 689   1 return set_host_ipvfuture(rv->str);
680   } 690   }
681   } 691   }
HITCBC 682   352 else if(s.size() >= 7) // "0.0.0.0" 692   352 else if(s.size() >= 7) // "0.0.0.0"
683   { 693   {
684   // IPv4-address 694   // IPv4-address
HITCBC 685   348 auto rv = parse_ipv4_address(s); 695   348 auto rv = parse_ipv4_address(s);
HITCBC 686   348 if(rv) 696   348 if(rv)
HITCBC 687   189 return set_host_ipv4(*rv); 697   189 return set_host_ipv4(*rv);
688   } 698   }
689   699  
690   // reg-name 700   // reg-name
HITCBC 691   163 op_t op(*this, &s); 701   163 op_t op(*this, &s);
HITCBC 692   163 encoding_opts opt; 702   163 encoding_opts opt;
HITCBC 693   163 auto const n = encoded_size( 703   163 auto const n = encoded_size(
694   s, detail::host_chars, opt); 704   s, detail::host_chars, opt);
HITCBC 695   163 auto dest = set_host_impl(n, op); 705   163 auto dest = set_host_impl(n, op);
HITCBC 696   163 encode( 706   163 encode(
697   dest, 707   dest,
HITCBC 698   163 impl_.get(id_path).data() - dest, 708   163 impl_.get(id_path).data() - dest,
699   s, 709   s,
700   detail::host_chars, 710   detail::host_chars,
701   opt); 711   opt);
HITCBC 702   163 impl_.decoded_[id_host] = 712   163 impl_.decoded_[id_host] =
HITCBC 703   163 detail::to_size_type(s.size()); 713   163 detail::to_size_type(s.size());
HITCBC 704   163 impl_.host_type_ = 714   163 impl_.host_type_ =
705   urls::host_type::name; 715   urls::host_type::name;
HITCBC 706   163 return *this; 716   163 return *this;
HITCBC 707   163 } 717   163 }
708   718  
709   // set host part from encoded text 719   // set host part from encoded text
710   inline 720   inline
711   url_base& 721   url_base&
HITCBC 712   221 url_base:: 722   221 url_base::
713   set_encoded_host( 723   set_encoded_host(
714   pct_string_view s) 724   pct_string_view s)
715   { 725   {
HITCBC 716   221 if( s.size() > 2 && 726   221 if( s.size() > 2 &&
HITCBC 717   241 s.front() == '[' && 727   241 s.front() == '[' &&
HITCBC 718   20 s.back() == ']') 728   20 s.back() == ']')
719   { 729   {
720   // IP-literal 730   // IP-literal
HITCBC 721   20 if (s[1] != 'v') 731   20 if (s[1] != 'v')
722   { 732   {
723   // IPv6-address 733   // IPv6-address
HITCBC 724   19 auto innersv = s.substr(1, s.size() - 2); 734   19 auto innersv = s.substr(1, s.size() - 2);
HITCBC 725   19 auto innerit = innersv.begin(); 735   19 auto innerit = innersv.begin();
HITCBC 726   19 auto endit = innersv.end(); 736   19 auto endit = innersv.end();
HITCBC 727   19 auto rv = grammar::parse( 737   19 auto rv = grammar::parse(
728   innerit, 738   innerit,
729   endit, 739   endit,
730   ipv6_address_rule); 740   ipv6_address_rule);
HITCBC 731   19 if(rv) 741   19 if(rv)
732   { 742   {
HITCBC 733   11 if (innerit == endit) 743   11 if (innerit == endit)
734   { 744   {
HITCBC 735   5 set_host_ipv6_and_encoded_zone_id(*rv, {}); 745   5 set_host_ipv6_and_encoded_zone_id(*rv, {});
HITCBC 736   9 return *this; 746   9 return *this;
737   } 747   }
738   // IPv6addrz: https://datatracker.ietf.org/doc/html/rfc6874 748   // IPv6addrz: https://datatracker.ietf.org/doc/html/rfc6874
HITCBC 739   6 auto chars_left = endit - innerit; 749   6 auto chars_left = endit - innerit;
HITCBC 740   10 if (chars_left >= 3 && 750   10 if (chars_left >= 3 &&
HITCBC 741   4 *innerit++ == '%' && 751   4 *innerit++ == '%' &&
HITCBC 742   14 *innerit++ == '2' && 752   14 *innerit++ == '2' &&
HITCBC 743   4 *innerit++ == '5') 753   4 *innerit++ == '5')
744   { 754   {
HITCBC 745   4 auto const nz = std::size_t(chars_left - 3); 755   4 auto const nz = std::size_t(chars_left - 3);
HITCBC 746   4 core::string_view zone_id_str = {&*innerit, std::size_t(chars_left - 3)}; 756   4 core::string_view zone_id_str = {&*innerit, std::size_t(chars_left - 3)};
HITCBC 747   4 std::size_t dnz = detail::decode_bytes_unsafe(zone_id_str); 757   4 std::size_t dnz = detail::decode_bytes_unsafe(zone_id_str);
HITCBC 748   4 pct_string_view zone_id_pct = make_pct_string_view_unsafe(innerit, nz, dnz); 758   4 pct_string_view zone_id_pct = make_pct_string_view_unsafe(innerit, nz, dnz);
HITCBC 749   4 set_host_ipv6_and_encoded_zone_id(*rv, zone_id_pct); 759   4 set_host_ipv6_and_encoded_zone_id(*rv, zone_id_pct);
HITCBC 750   4 return *this; 760   4 return *this;
751   } 761   }
752   } 762   }
753   } 763   }
754   else 764   else
755   { 765   {
756   // IPvFuture 766   // IPvFuture
HITCBC 757   1 auto rv = grammar::parse( 767   1 auto rv = grammar::parse(
HITCBC 758   1 s.substr(1, s.size() - 2), 768   1 s.substr(1, s.size() - 2),
759   detail::ipvfuture_rule); 769   detail::ipvfuture_rule);
HITCBC 760   1 if(rv) 770   1 if(rv)
HITCBC 761   1 return set_host_ipvfuture(rv->str); 771   1 return set_host_ipvfuture(rv->str);
762   } 772   }
763   } 773   }
HITCBC 764   201 else if(s.size() >= 7) // "0.0.0.0" 774   201 else if(s.size() >= 7) // "0.0.0.0"
765   { 775   {
766   // IPv4-address 776   // IPv4-address
HITCBC 767   74 auto rv = parse_ipv4_address(s); 777   74 auto rv = parse_ipv4_address(s);
HITCBC 768   74 if(rv) 778   74 if(rv)
HITCBC 769   5 return set_host_ipv4(*rv); 779   5 return set_host_ipv4(*rv);
770   } 780   }
771   781  
772   // reg-name 782   // reg-name
HITCBC 773   206 op_t op(*this, &detail::ref(s)); 783   206 op_t op(*this, &detail::ref(s));
HITCBC 774   206 auto const n = detail::re_encoded_size_unsafe( 784   206 auto const n = detail::re_encoded_size_unsafe(
775   s, detail::host_chars); 785   s, detail::host_chars);
HITCBC 776   206 auto dest = set_host_impl(n, op); 786   206 auto dest = set_host_impl(n, op);
HITCBC 777   206 impl_.decoded_[id_host] = 787   206 impl_.decoded_[id_host] =
HITCBC 778   412 detail::to_size_type(detail::re_encode_unsafe( 788   412 detail::to_size_type(detail::re_encode_unsafe(
779   dest, 789   dest,
HITCBC 780   206 impl_.get(id_path).data(), 790   206 impl_.get(id_path).data(),
781   s, 791   s,
782   detail::host_chars)); 792   detail::host_chars));
HITCBC 783   206 BOOST_ASSERT(impl_.decoded_[id_host] == 793   206 BOOST_ASSERT(impl_.decoded_[id_host] ==
784   s.decoded_size()); 794   s.decoded_size());
HITCBC 785   206 impl_.host_type_ = 795   206 impl_.host_type_ =
786   urls::host_type::name; 796   urls::host_type::name;
HITCBC 787   206 return *this; 797   206 return *this;
HITCBC 788   206 } 798   206 }
789   799  
790   inline 800   inline
791   url_base& 801   url_base&
HITCBC 792   13 url_base:: 802   13 url_base::
793   set_host_address( 803   set_host_address(
794   core::string_view s) 804   core::string_view s)
795   { 805   {
HITCBC 796   13 if (!s.empty()) 806   13 if (!s.empty())
797   { 807   {
798   // IP-literal 808   // IP-literal
HITCBC 799   12 if (s[0] != 'v') 809   12 if (s[0] != 'v')
800   { 810   {
801   // IPv6-address 811   // IPv6-address
HITCBC 802   11 auto innerit = s.begin(); 812   11 auto innerit = s.begin();
HITCBC 803   11 auto endit = s.end(); 813   11 auto endit = s.end();
HITCBC 804   11 auto rv = grammar::parse( 814   11 auto rv = grammar::parse(
805   innerit, 815   innerit,
806   endit, 816   endit,
807   ipv6_address_rule); 817   ipv6_address_rule);
HITCBC 808   11 if(rv) 818   11 if(rv)
809   { 819   {
HITCBC 810   5 if (innerit == endit) 820   5 if (innerit == endit)
811   { 821   {
HITCBC 812   1 set_host_ipv6_and_encoded_zone_id(*rv, {}); 822   1 set_host_ipv6_and_encoded_zone_id(*rv, {});
HITCBC 813   5 return *this; 823   5 return *this;
814   } 824   }
815   // IPv6addrz: https://datatracker.ietf.org/doc/html/rfc6874 825   // IPv6addrz: https://datatracker.ietf.org/doc/html/rfc6874
HITCBC 816   4 auto chars_left = endit - innerit; 826   4 auto chars_left = endit - innerit;
HITCBC 817   8 if (chars_left >= 2 && 827   8 if (chars_left >= 2 &&
HITCBC 818   4 *innerit++ == '%') 828   4 *innerit++ == '%')
819   { 829   {
HITCBC 820   4 core::string_view zone_id_str = {&*innerit, std::size_t(chars_left - 1)}; 830   4 core::string_view zone_id_str = {&*innerit, std::size_t(chars_left - 1)};
HITCBC 821   4 set_host_ipv6_and_zone_id(*rv, zone_id_str); 831   4 set_host_ipv6_and_zone_id(*rv, zone_id_str);
HITCBC 822   4 return *this; 832   4 return *this;
823   } 833   }
824   } 834   }
825   } 835   }
826   836  
827   // IPvFuture 837   // IPvFuture
HITCBC 828   7 auto rv = grammar::parse(s, detail::ipvfuture_rule); 838   7 auto rv = grammar::parse(s, detail::ipvfuture_rule);
HITCBC 829   7 if(rv) 839   7 if(rv)
HITCBC 830   1 return set_host_ipvfuture(rv->str); 840   1 return set_host_ipvfuture(rv->str);
831   841  
HITCBC 832   6 if(s.size() >= 7) // "0.0.0.0" 842   6 if(s.size() >= 7) // "0.0.0.0"
833   { 843   {
834   // IPv4-address 844   // IPv4-address
HITCBC 835   5 auto rv2 = parse_ipv4_address(s); 845   5 auto rv2 = parse_ipv4_address(s);
HITCBC 836   5 if(rv2) 846   5 if(rv2)
HITCBC 837   2 return set_host_ipv4(*rv2); 847   2 return set_host_ipv4(*rv2);
838   } 848   }
839   } 849   }
840   850  
841   // reg-name 851   // reg-name
HITCBC 842   5 op_t op(*this, &s); 852   5 op_t op(*this, &s);
HITCBC 843   5 encoding_opts opt; 853   5 encoding_opts opt;
HITCBC 844   5 auto const n = encoded_size( 854   5 auto const n = encoded_size(
845   s, detail::host_chars, opt); 855   s, detail::host_chars, opt);
HITCBC 846   5 auto dest = set_host_impl(n, op); 856   5 auto dest = set_host_impl(n, op);
HITCBC 847   5 encode( 857   5 encode(
848   dest, 858   dest,
HITCBC 849   5 impl_.get(id_path).data() - dest, 859   5 impl_.get(id_path).data() - dest,
850   s, 860   s,
851   detail::host_chars, 861   detail::host_chars,
852   opt); 862   opt);
HITCBC 853   5 impl_.decoded_[id_host] = 863   5 impl_.decoded_[id_host] =
HITCBC 854   5 detail::to_size_type(s.size()); 864   5 detail::to_size_type(s.size());
HITCBC 855   5 impl_.host_type_ = 865   5 impl_.host_type_ =
856   urls::host_type::name; 866   urls::host_type::name;
HITCBC 857   5 return *this; 867   5 return *this;
HITCBC 858   5 } 868   5 }
859   869  
860   inline 870   inline
861   url_base& 871   url_base&
HITCBC 862   11 url_base:: 872   11 url_base::
863   set_encoded_host_address( 873   set_encoded_host_address(
864   pct_string_view s) 874   pct_string_view s)
865   { 875   {
HITCBC 866   11 if( !s.empty() ) 876   11 if( !s.empty() )
867   { 877   {
868   // IP-literal 878   // IP-literal
HITCBC 869   10 if (s[0] != 'v') 879   10 if (s[0] != 'v')
870   { 880   {
871   // IPv6-address 881   // IPv6-address
HITCBC 872   9 auto innerit = s.begin(); 882   9 auto innerit = s.begin();
HITCBC 873   9 auto endit = s.end(); 883   9 auto endit = s.end();
HITCBC 874   9 auto rv = grammar::parse( 884   9 auto rv = grammar::parse(
875   innerit, 885   innerit,
876   endit, 886   endit,
877   ipv6_address_rule); 887   ipv6_address_rule);
HITCBC 878   9 if(rv) 888   9 if(rv)
879   { 889   {
HITCBC 880   5 if (innerit == endit) 890   5 if (innerit == endit)
881   { 891   {
HITCBC 882   1 set_host_ipv6_and_encoded_zone_id(*rv, {}); 892   1 set_host_ipv6_and_encoded_zone_id(*rv, {});
HITCBC 883   6 return *this; 893   6 return *this;
884   } 894   }
885   // IPv6addrz: https://datatracker.ietf.org/doc/html/rfc6874 895   // IPv6addrz: https://datatracker.ietf.org/doc/html/rfc6874
HITCBC 886   4 auto chars_left = endit - innerit; 896   4 auto chars_left = endit - innerit;
HITCBC 887   8 if (chars_left >= 3 && 897   8 if (chars_left >= 3 &&
HITCBC 888   4 *innerit++ == '%' && 898   4 *innerit++ == '%' &&
HITCBC 889   12 *innerit++ == '2' && 899   12 *innerit++ == '2' &&
HITCBC 890   4 *innerit++ == '5') 900   4 *innerit++ == '5')
891   { 901   {
HITCBC 892   4 auto const nz = std::size_t(chars_left - 3); 902   4 auto const nz = std::size_t(chars_left - 3);
HITCBC 893   4 core::string_view zone_id_str = {&*innerit, std::size_t(chars_left - 3)}; 903   4 core::string_view zone_id_str = {&*innerit, std::size_t(chars_left - 3)};
HITCBC 894   4 std::size_t dnz = detail::decode_bytes_unsafe(zone_id_str); 904   4 std::size_t dnz = detail::decode_bytes_unsafe(zone_id_str);
HITCBC 895   4 pct_string_view zone_id_pct = make_pct_string_view_unsafe(innerit, nz, dnz); 905   4 pct_string_view zone_id_pct = make_pct_string_view_unsafe(innerit, nz, dnz);
HITCBC 896   4 set_host_ipv6_and_encoded_zone_id(*rv, zone_id_pct); 906   4 set_host_ipv6_and_encoded_zone_id(*rv, zone_id_pct);
HITCBC 897   4 return *this; 907   4 return *this;
898   } 908   }
899   } 909   }
900   910  
HITCBC 901   4 if(s.size() >= 7) // "0.0.0.0" 911   4 if(s.size() >= 7) // "0.0.0.0"
902   { 912   {
903   // IPv4-address 913   // IPv4-address
HITCBC 904   3 auto rv2 = parse_ipv4_address(s); 914   3 auto rv2 = parse_ipv4_address(s);
HITCBC 905   3 if(rv2) 915   3 if(rv2)
HITCBC 906   1 return set_host_ipv4(*rv2); 916   1 return set_host_ipv4(*rv2);
907   } 917   }
908   } 918   }
909   else 919   else
910   { 920   {
911   // IPvFuture 921   // IPvFuture
HITCBC 912   1 auto rv = grammar::parse( 922   1 auto rv = grammar::parse(
913   s, detail::ipvfuture_rule); 923   s, detail::ipvfuture_rule);
HITCBC 914   1 if(rv) 924   1 if(rv)
HITCBC 915   1 return set_host_ipvfuture(rv->str); 925   1 return set_host_ipvfuture(rv->str);
916   } 926   }
917   } 927   }
918   928  
919   // reg-name 929   // reg-name
HITCBC 920   4 op_t op(*this, &detail::ref(s)); 930   4 op_t op(*this, &detail::ref(s));
HITCBC 921   4 auto const n = detail::re_encoded_size_unsafe( 931   4 auto const n = detail::re_encoded_size_unsafe(
922   s, detail::host_chars); 932   s, detail::host_chars);
HITCBC 923   4 auto dest = set_host_impl(n, op); 933   4 auto dest = set_host_impl(n, op);
HITCBC 924   4 impl_.decoded_[id_host] = 934   4 impl_.decoded_[id_host] =
HITCBC 925   8 detail::to_size_type(detail::re_encode_unsafe( 935   8 detail::to_size_type(detail::re_encode_unsafe(
926   dest, 936   dest,
HITCBC 927   4 impl_.get(id_path).data(), 937   4 impl_.get(id_path).data(),
928   s, 938   s,
929   detail::host_chars)); 939   detail::host_chars));
HITCBC 930   4 BOOST_ASSERT(impl_.decoded_[id_host] == 940   4 BOOST_ASSERT(impl_.decoded_[id_host] ==
931   s.decoded_size()); 941   s.decoded_size());
HITCBC 932   4 impl_.host_type_ = 942   4 impl_.host_type_ =
933   urls::host_type::name; 943   urls::host_type::name;
HITCBC 934   4 return *this; 944   4 return *this;
HITCBC 935   4 } 945   4 }
936   946  
937   inline 947   inline
938   url_base& 948   url_base&
HITCBC 939   203 url_base:: 949   203 url_base::
940   set_host_ipv4( 950   set_host_ipv4(
941   ipv4_address const& addr) 951   ipv4_address const& addr)
942   { 952   {
HITCBC 943   203 op_t op(*this); 953   203 op_t op(*this);
944   char buf[urls::ipv4_address::max_str_len]; 954   char buf[urls::ipv4_address::max_str_len];
HITCBC 945   203 auto s = addr.to_buffer(buf, sizeof(buf)); 955   203 auto s = addr.to_buffer(buf, sizeof(buf));
HITCBC 946   203 auto dest = set_host_impl(s.size(), op); 956   203 auto dest = set_host_impl(s.size(), op);
HITCBC 947   203 std::memcpy(dest, s.data(), s.size()); 957   203 std::memcpy(dest, s.data(), s.size());
HITCBC 948   203 impl_.decoded_[id_host] = 958   203 impl_.decoded_[id_host] =
HITCBC 949   203 detail::to_size_type(impl_.len(id_host)); 959   203 detail::to_size_type(impl_.len(id_host));
HITCBC 950   203 impl_.host_type_ = urls::host_type::ipv4; 960   203 impl_.host_type_ = urls::host_type::ipv4;
HITCBC 951   203 auto bytes = addr.to_bytes(); 961   203 auto bytes = addr.to_bytes();
HITCBC 952   203 std::memcpy( 962   203 std::memcpy(
HITCBC 953   203 impl_.ip_addr_, 963   203 impl_.ip_addr_,
HITCBC 954   203 bytes.data(), 964   203 bytes.data(),
955   bytes.size()); 965   bytes.size());
HITCBC 956   203 return *this; 966   203 return *this;
HITCBC 957   203 } 967   203 }
958   968  
959   inline 969   inline
960   url_base& 970   url_base&
HITCBC 961   5 url_base:: 971   5 url_base::
962   set_host_ipv6( 972   set_host_ipv6(
963   ipv6_address const& addr) 973   ipv6_address const& addr)
964   { 974   {
HITCBC 965   5 set_host_ipv6_and_encoded_zone_id(addr, encoded_zone_id()); 975   5 set_host_ipv6_and_encoded_zone_id(addr, encoded_zone_id());
HITCBC 966   5 return *this; 976   5 return *this;
967   } 977   }
968   978  
969   inline 979   inline
970   url_base& 980   url_base&
HITCBC 971   6 url_base:: 981   6 url_base::
972   set_zone_id(core::string_view s) 982   set_zone_id(core::string_view s)
973   { 983   {
HITCBC 974   6 set_host_ipv6_and_zone_id(host_ipv6_address(), s); 984   6 set_host_ipv6_and_zone_id(host_ipv6_address(), s);
HITCBC 975   6 return *this; 985   6 return *this;
976   } 986   }
977   987  
978   inline 988   inline
979   url_base& 989   url_base&
HITCBC 980   6 url_base:: 990   6 url_base::
981   set_encoded_zone_id(pct_string_view s) 991   set_encoded_zone_id(pct_string_view s)
982   { 992   {
HITCBC 983   6 set_host_ipv6_and_encoded_zone_id(host_ipv6_address(), s); 993   6 set_host_ipv6_and_encoded_zone_id(host_ipv6_address(), s);
HITCBC 984   6 return *this; 994   6 return *this;
985   } 995   }
986   996  
987   inline 997   inline
988   void 998   void
HITCBC 989   14 url_base:: 999   14 url_base::
990   set_host_ipv6_and_zone_id( 1000   set_host_ipv6_and_zone_id(
991   ipv6_address const& addr, 1001   ipv6_address const& addr,
992   core::string_view zone_id) 1002   core::string_view zone_id)
993   { 1003   {
HITCBC 994   14 op_t op(*this, &zone_id); 1004   14 op_t op(*this, &zone_id);
995   char ipv6_str_buf[urls::ipv6_address::max_str_len]; 1005   char ipv6_str_buf[urls::ipv6_address::max_str_len];
HITCBC 996   14 auto ipv6_str = addr.to_buffer(ipv6_str_buf, sizeof(ipv6_str_buf)); 1006   14 auto ipv6_str = addr.to_buffer(ipv6_str_buf, sizeof(ipv6_str_buf));
HITCBC 997   14 bool const has_zone_id = !zone_id.empty(); 1007   14 bool const has_zone_id = !zone_id.empty();
HITCBC 998   14 encoding_opts opt; 1008   14 encoding_opts opt;
HITCBC 999   14 auto const ipn = ipv6_str.size(); 1009   14 auto const ipn = ipv6_str.size();
HITCBC 1000   14 auto const zn = encoded_size(zone_id, unreserved_chars, opt); 1010   14 auto const zn = encoded_size(zone_id, unreserved_chars, opt);
HITCBC 1001   14 auto const n = ipn + 2 + has_zone_id * (3 + zn); 1011   14 auto const n = ipn + 2 + has_zone_id * (3 + zn);
HITCBC 1002   14 auto dest = set_host_impl(n, op); 1012   14 auto dest = set_host_impl(n, op);
HITCBC 1003   14 *dest++ = '['; 1013   14 *dest++ = '[';
HITCBC 1004   14 std::memcpy(dest, ipv6_str.data(), ipn); 1014   14 std::memcpy(dest, ipv6_str.data(), ipn);
HITCBC 1005   14 dest += ipn; 1015   14 dest += ipn;
HITCBC 1006   14 if (has_zone_id) 1016   14 if (has_zone_id)
1007   { 1017   {
HITCBC 1008   14 *dest++ = '%'; 1018   14 *dest++ = '%';
HITCBC 1009   14 *dest++ = '2'; 1019   14 *dest++ = '2';
HITCBC 1010   14 *dest++ = '5'; 1020   14 *dest++ = '5';
HITCBC 1011   14 encode(dest, zn, zone_id, unreserved_chars, opt); 1021   14 encode(dest, zn, zone_id, unreserved_chars, opt);
HITCBC 1012   14 dest += zn; 1022   14 dest += zn;
1013   } 1023   }
HITCBC 1014   14 *dest++ = ']'; 1024   14 *dest++ = ']';
1015   // ipn + |"["| + |"]"| + (has_zone_id ? |"%"| + zn : 0) 1025   // ipn + |"["| + |"]"| + (has_zone_id ? |"%"| + zn : 0)
HITCBC 1016   28 impl_.decoded_[id_host] = detail::to_size_type( 1026   28 impl_.decoded_[id_host] = detail::to_size_type(
HITCBC 1017   14 ipn + 2 + has_zone_id * (1 + zone_id.size())); 1027   14 ipn + 2 + has_zone_id * (1 + zone_id.size()));
HITCBC 1018   14 impl_.host_type_ = urls::host_type::ipv6; 1028   14 impl_.host_type_ = urls::host_type::ipv6;
HITCBC 1019   14 auto bytes = addr.to_bytes(); 1029   14 auto bytes = addr.to_bytes();
HITCBC 1020   14 std::memcpy( 1030   14 std::memcpy(
HITCBC 1021   14 impl_.ip_addr_, 1031   14 impl_.ip_addr_,
HITCBC 1022   14 bytes.data(), 1032   14 bytes.data(),
1023   bytes.size()); 1033   bytes.size());
HITCBC 1024   14 } 1034   14 }
1025   1035  
1026   inline 1036   inline
1027   void 1037   void
HITCBC 1028   73 url_base:: 1038   73 url_base::
1029   set_host_ipv6_and_encoded_zone_id( 1039   set_host_ipv6_and_encoded_zone_id(
1030   ipv6_address const& addr, 1040   ipv6_address const& addr,
1031   pct_string_view zone_id) 1041   pct_string_view zone_id)
1032   { 1042   {
HITCBC 1033   73 op_t op(*this, &detail::ref(zone_id)); 1043   73 op_t op(*this, &detail::ref(zone_id));
1034   char ipv6_str_buf[urls::ipv6_address::max_str_len]; 1044   char ipv6_str_buf[urls::ipv6_address::max_str_len];
HITCBC 1035   73 auto ipv6_str = addr.to_buffer(ipv6_str_buf, sizeof(ipv6_str_buf)); 1045   73 auto ipv6_str = addr.to_buffer(ipv6_str_buf, sizeof(ipv6_str_buf));
HITCBC 1036   73 bool const has_zone_id = !zone_id.empty(); 1046   73 bool const has_zone_id = !zone_id.empty();
HITCBC 1037   73 auto const ipn = ipv6_str.size(); 1047   73 auto const ipn = ipv6_str.size();
HITCBC 1038   73 auto const zn = detail::re_encoded_size_unsafe(zone_id, unreserved_chars); 1048   73 auto const zn = detail::re_encoded_size_unsafe(zone_id, unreserved_chars);
HITCBC 1039   73 auto const n = ipn + 2 + has_zone_id * (3 + zn); 1049   73 auto const n = ipn + 2 + has_zone_id * (3 + zn);
HITCBC 1040   73 auto dest = set_host_impl(n, op); 1050   73 auto dest = set_host_impl(n, op);
HITCBC 1041   73 *dest++ = '['; 1051   73 *dest++ = '[';
HITCBC 1042   73 std::memcpy(dest, ipv6_str.data(), ipn); 1052   73 std::memcpy(dest, ipv6_str.data(), ipn);
HITCBC 1043   73 dest += ipn; 1053   73 dest += ipn;
HITCBC 1044   73 std::size_t dzn = 0; 1054   73 std::size_t dzn = 0;
HITCBC 1045   73 if (has_zone_id) 1055   73 if (has_zone_id)
1046   { 1056   {
HITCBC 1047   15 *dest++ = '%'; 1057   15 *dest++ = '%';
HITCBC 1048   15 *dest++ = '2'; 1058   15 *dest++ = '2';
HITCBC 1049   15 *dest++ = '5'; 1059   15 *dest++ = '5';
HITCBC 1050   15 dzn = detail::re_encode_unsafe(dest, dest + zn, zone_id, unreserved_chars); 1060   15 dzn = detail::re_encode_unsafe(dest, dest + zn, zone_id, unreserved_chars);
1051   } 1061   }
HITCBC 1052   73 *dest++ = ']'; 1062   73 *dest++ = ']';
1053   // ipn + |"["| + |"]"| + (has_zone_id ? |"%"| + zn : 0) 1063   // ipn + |"["| + |"]"| + (has_zone_id ? |"%"| + zn : 0)
HITCBC 1054   146 impl_.decoded_[id_host] = detail::to_size_type( 1064   146 impl_.decoded_[id_host] = detail::to_size_type(
HITCBC 1055   73 ipn + 2 + has_zone_id * (1 + dzn)); 1065   73 ipn + 2 + has_zone_id * (1 + dzn));
HITCBC 1056   73 impl_.host_type_ = urls::host_type::ipv6; 1066   73 impl_.host_type_ = urls::host_type::ipv6;
HITCBC 1057   73 auto bytes = addr.to_bytes(); 1067   73 auto bytes = addr.to_bytes();
HITCBC 1058   73 std::memcpy( 1068   73 std::memcpy(
HITCBC 1059   73 impl_.ip_addr_, 1069   73 impl_.ip_addr_,
HITCBC 1060   73 bytes.data(), 1070   73 bytes.data(),
1061   bytes.size()); 1071   bytes.size());
HITCBC 1062   73 } 1072   73 }
1063   1073  
1064   inline 1074   inline
1065   url_base& 1075   url_base&
HITCBC 1066   7 url_base:: 1076   7 url_base::
1067   set_host_ipvfuture( 1077   set_host_ipvfuture(
1068   core::string_view s) 1078   core::string_view s)
1069   { 1079   {
HITCBC 1070   7 op_t op(*this, &s); 1080   7 op_t op(*this, &s);
1071   // validate 1081   // validate
HITCBC 1072   8 grammar::parse(s, 1082   8 grammar::parse(s,
1073   detail::ipvfuture_rule 1083   detail::ipvfuture_rule
HITCBC 1074   8 ).value(BOOST_URL_POS); 1084   8 ).value(BOOST_URL_POS);
HITCBC 1075   6 auto dest = set_host_impl( 1085   6 auto dest = set_host_impl(
HITCBC 1076   6 s.size() + 2, op); 1086   6 s.size() + 2, op);
HITCBC 1077   6 *dest++ = '['; 1087   6 *dest++ = '[';
HITCBC 1078   6 dest += s.copy(dest, s.size()); 1088   6 dest += s.copy(dest, s.size());
HITCBC 1079   6 *dest = ']'; 1089   6 *dest = ']';
HITCBC 1080   6 impl_.host_type_ = 1090   6 impl_.host_type_ =
1081   urls::host_type::ipvfuture; 1091   urls::host_type::ipvfuture;
HITCBC 1082   6 impl_.decoded_[id_host] = 1092   6 impl_.decoded_[id_host] =
HITCBC 1083   6 detail::to_size_type(s.size() + 2); 1093   6 detail::to_size_type(s.size() + 2);
HITCBC 1084   6 return *this; 1094   6 return *this;
HITCBC 1085   7 } 1095   7 }
1086   1096  
1087   inline 1097   inline
1088   url_base& 1098   url_base&
HITCBC 1089   4 url_base:: 1099   4 url_base::
1090   set_host_name( 1100   set_host_name(
1091   core::string_view s) 1101   core::string_view s)
1092   { 1102   {
HITCBC 1093   4 bool is_ipv4 = false; 1103   4 bool is_ipv4 = false;
HITCBC 1094   4 if(s.size() >= 7) // "0.0.0.0" 1104   4 if(s.size() >= 7) // "0.0.0.0"
1095   { 1105   {
1096   // IPv4-address 1106   // IPv4-address
HITCBC 1097   3 if(parse_ipv4_address(s).has_value()) 1107   3 if(parse_ipv4_address(s).has_value())
HITCBC 1098   1 is_ipv4 = true; 1108   1 is_ipv4 = true;
1099   } 1109   }
HITCBC 1100   4 auto allowed = detail::host_chars; 1110   4 auto allowed = detail::host_chars;
HITCBC 1101   4 if(is_ipv4) 1111   4 if(is_ipv4)
HITCBC 1102   1 allowed = allowed - '.'; 1112   1 allowed = allowed - '.';
1103   1113  
HITCBC 1104   4 op_t op(*this, &s); 1114   4 op_t op(*this, &s);
HITCBC 1105   4 encoding_opts opt; 1115   4 encoding_opts opt;
HITCBC 1106   4 auto const n = encoded_size( 1116   4 auto const n = encoded_size(
1107   s, allowed, opt); 1117   s, allowed, opt);
HITCBC 1108   4 auto dest = set_host_impl(n, op); 1118   4 auto dest = set_host_impl(n, op);
HITCBC 1109   4 encode_unsafe( 1119   4 encode_unsafe(
1110   dest, 1120   dest,
1111   n, 1121   n,
1112   s, 1122   s,
1113   allowed, 1123   allowed,
1114   opt); 1124   opt);
HITCBC 1115   4 impl_.host_type_ = 1125   4 impl_.host_type_ =
1116   urls::host_type::name; 1126   urls::host_type::name;
HITCBC 1117   4 impl_.decoded_[id_host] = 1127   4 impl_.decoded_[id_host] =
HITCBC 1118   4 detail::to_size_type(s.size()); 1128   4 detail::to_size_type(s.size());
HITCBC 1119   4 return *this; 1129   4 return *this;
HITCBC 1120   4 } 1130   4 }
1121   1131  
1122   inline 1132   inline
1123   url_base& 1133   url_base&
HITCBC 1124   4 url_base:: 1134   4 url_base::
1125   set_encoded_host_name( 1135   set_encoded_host_name(
1126   pct_string_view s) 1136   pct_string_view s)
1127   { 1137   {
HITCBC 1128   4 bool is_ipv4 = false; 1138   4 bool is_ipv4 = false;
HITCBC 1129   4 if(s.size() >= 7) // "0.0.0.0" 1139   4 if(s.size() >= 7) // "0.0.0.0"
1130   { 1140   {
1131   // IPv4-address 1141   // IPv4-address
HITCBC 1132   3 if(parse_ipv4_address(s).has_value()) 1142   3 if(parse_ipv4_address(s).has_value())
HITCBC 1133   1 is_ipv4 = true; 1143   1 is_ipv4 = true;
1134   } 1144   }
HITCBC 1135   4 auto allowed = detail::host_chars; 1145   4 auto allowed = detail::host_chars;
HITCBC 1136   4 if(is_ipv4) 1146   4 if(is_ipv4)
HITCBC 1137   1 allowed = allowed - '.'; 1147   1 allowed = allowed - '.';
1138   1148  
HITCBC 1139   4 op_t op(*this, &detail::ref(s)); 1149   4 op_t op(*this, &detail::ref(s));
HITCBC 1140   4 auto const n = detail::re_encoded_size_unsafe( 1150   4 auto const n = detail::re_encoded_size_unsafe(
1141   s, allowed); 1151   s, allowed);
HITCBC 1142   4 auto dest = set_host_impl(n, op); 1152   4 auto dest = set_host_impl(n, op);
HITCBC 1143   4 impl_.decoded_[id_host] = 1153   4 impl_.decoded_[id_host] =
HITCBC 1144   8 detail::to_size_type(detail::re_encode_unsafe( 1154   8 detail::to_size_type(detail::re_encode_unsafe(
1145   dest, 1155   dest,
HITCBC 1146   4 dest + n, 1156   4 dest + n,
1147   s, 1157   s,
1148   allowed)); 1158   allowed));
HITCBC 1149   4 BOOST_ASSERT( 1159   4 BOOST_ASSERT(
1150   impl_.decoded_[id_host] == 1160   impl_.decoded_[id_host] ==
1151   s.decoded_size()); 1161   s.decoded_size());
HITCBC 1152   4 impl_.host_type_ = 1162   4 impl_.host_type_ =
1153   urls::host_type::name; 1163   urls::host_type::name;
HITCBC 1154   4 return *this; 1164   4 return *this;
HITCBC 1155   4 } 1165   4 }
1156   1166  
1157   //------------------------------------------------ 1167   //------------------------------------------------
1158   1168  
1159   inline 1169   inline
1160   url_base& 1170   url_base&
HITCBC 1161   215 url_base:: 1171   215 url_base::
1162   set_port_number( 1172   set_port_number(
1163   std::uint16_t n) 1173   std::uint16_t n)
1164   { 1174   {
HITCBC 1165   215 op_t op(*this); 1175   215 op_t op(*this);
1166   auto s = 1176   auto s =
HITCBC 1167   215 detail::make_printed(n); 1177   215 detail::make_printed(n);
HITCBC 1168   215 auto dest = set_port_impl( 1178   215 auto dest = set_port_impl(
HITCBC 1169   215 s.string().size(), op); 1179   215 s.string().size(), op);
HITCBC 1170   215 std::memcpy( 1180   215 std::memcpy(
HITCBC 1171   215 dest, s.string().data(), 1181   215 dest, s.string().data(),
HITCBC 1172   215 s.string().size()); 1182   215 s.string().size());
HITCBC 1173   215 impl_.port_number_ = n; 1183   215 impl_.port_number_ = n;
HITCBC 1174   215 return *this; 1184   215 return *this;
HITCBC 1175   215 } 1185   215 }
1176   1186  
1177   inline 1187   inline
1178   url_base& 1188   url_base&
HITCBC 1179   389 url_base:: 1189   389 url_base::
1180   set_port( 1190   set_port(
1181   core::string_view s) 1191   core::string_view s)
1182   { 1192   {
HITCBC 1183   389 op_t op(*this, &s); 1193   389 op_t op(*this, &s);
HITCBC 1184   410 auto t = grammar::parse(s, 1194   410 auto t = grammar::parse(s,
HITCBC 1185   21 detail::port_rule{} 1195   21 detail::port_rule{}
HITCBC 1186   389 ).value(BOOST_URL_POS); 1196   389 ).value(BOOST_URL_POS);
1187   auto dest = 1197   auto dest =
HITCBC 1188   368 set_port_impl(t.str.size(), op); 1198   368 set_port_impl(t.str.size(), op);
HITCBC 1189   368 std::memcpy(dest, 1199   368 std::memcpy(dest,
HITCBC 1190   368 t.str.data(), t.str.size()); 1200   368 t.str.data(), t.str.size());
HITCBC 1191   368 if(t.has_number) 1201   368 if(t.has_number)
HITCBC 1192   208 impl_.port_number_ = t.number; 1202   208 impl_.port_number_ = t.number;
1193   else 1203   else
HITCBC 1194   160 impl_.port_number_ = 0; 1204   160 impl_.port_number_ = 0;
HITCBC 1195   368 return *this; 1205   368 return *this;
HITCBC 1196   389 } 1206   389 }
1197   1207  
1198   inline 1208   inline
1199   url_base& 1209   url_base&
HITCBC 1200   206 url_base:: 1210   206 url_base::
1201   remove_port() noexcept 1211   remove_port() noexcept
1202   { 1212   {
HITCBC 1203   206 op_t op(*this); 1213   206 op_t op(*this);
HITCBC 1204   206 resize_impl(id_port, 0, op); 1214   206 resize_impl(id_port, 0, op);
HITCBC 1205   206 impl_.port_number_ = 0; 1215   206 impl_.port_number_ = 0;
HITCBC 1206   412 return *this; 1216   412 return *this;
HITCBC 1207   206 } 1217   206 }
1208   1218  
1209   //------------------------------------------------ 1219   //------------------------------------------------
1210   // 1220   //
1211   // Compound Fields 1221   // Compound Fields
1212   // 1222   //
1213   //------------------------------------------------ 1223   //------------------------------------------------
1214   1224  
1215   inline 1225   inline
1216   url_base& 1226   url_base&
HITCBC 1217   14 url_base:: 1227   14 url_base::
1218   remove_origin() 1228   remove_origin()
1219   { 1229   {
1220   // these two calls perform 2 memmoves instead of 1 1230   // these two calls perform 2 memmoves instead of 1
HITCBC 1221   14 remove_authority(); 1231   14 remove_authority();
HITCBC 1222   14 remove_scheme(); 1232   14 remove_scheme();
HITCBC 1223   14 return *this; 1233   14 return *this;
1224   } 1234   }
1225   1235  
1226   //------------------------------------------------ 1236   //------------------------------------------------
1227   // 1237   //
1228   // Path 1238   // Path
1229   // 1239   //
1230   //------------------------------------------------ 1240   //------------------------------------------------
1231   1241  
1232   inline 1242   inline
1233   bool 1243   bool
HITCBC 1234   52 url_base:: 1244   52 url_base::
1235   set_path_absolute( 1245   set_path_absolute(
1236   bool absolute) 1246   bool absolute)
1237   { 1247   {
HITCBC 1238   52 op_t op(*this); 1248   52 op_t op(*this);
1239   1249  
1240   // check if path empty 1250   // check if path empty
HITCBC 1241   52 if(impl_.len(id_path) == 0) 1251   52 if(impl_.len(id_path) == 0)
1242   { 1252   {
HITCBC 1243   40 if(! absolute) 1253   40 if(! absolute)
1244   { 1254   {
1245   // already not absolute 1255   // already not absolute
HITCBC 1246   34 return true; 1256   34 return true;
1247   } 1257   }
1248   1258  
1249   // add '/' 1259   // add '/'
HITCBC 1250   6 auto dest = resize_impl( 1260   6 auto dest = resize_impl(
1251   id_path, 1, op); 1261   id_path, 1, op);
HITCBC 1252   6 *dest = '/'; 1262   6 *dest = '/';
HITCBC 1253   6 ++impl_.decoded_[id_path]; 1263   6 ++impl_.decoded_[id_path];
HITCBC 1254   6 return true; 1264   6 return true;
1255   } 1265   }
1256   1266  
1257   // check if path absolute 1267   // check if path absolute
HITCBC 1258   12 if(s_[impl_.offset(id_path)] == '/') 1268   12 if(s_[impl_.offset(id_path)] == '/')
1259   { 1269   {
HITCBC 1260   9 if(absolute) 1270   9 if(absolute)
1261   { 1271   {
1262   // already absolute 1272   // already absolute
HITCBC 1263   2 return true; 1273   2 return true;
1264   } 1274   }
1265   1275  
HITCBC 1266   11 if( has_authority() && 1276   11 if( has_authority() &&
HITCBC 1267   4 impl_.len(id_path) > 1) 1277   4 impl_.len(id_path) > 1)
1268   { 1278   {
1269   // can't do it, paths are always 1279   // can't do it, paths are always
1270   // absolute when authority present! 1280   // absolute when authority present!
HITCBC 1271   2 return false; 1281   2 return false;
1272   } 1282   }
1273   1283  
HITCBC 1274   5 auto p = encoded_path(); 1284   5 auto p = encoded_path();
HITCBC 1275   5 auto pos = p.find_first_of(":/", 1); 1285   5 auto pos = p.find_first_of(":/", 1);
HITCBC 1276   6 if (pos != core::string_view::npos && 1286   6 if (pos != core::string_view::npos &&
HITCBC 1277   1 p[pos] == ':') 1287   1 p[pos] == ':')
1278   { 1288   {
1279   // prepend with . 1289   // prepend with .
HITCBC 1280   1 auto n = impl_.len(id_path); 1290   1 auto n = impl_.len(id_path);
HITCBC 1281   1 resize_impl(id_path, n + 1, op); 1291   1 resize_impl(id_path, n + 1, op);
HITCBC 1282   1 std::memmove( 1292   1 std::memmove(
HITCBC 1283   2 s_ + impl_.offset(id_path) + 1, 1293   2 s_ + impl_.offset(id_path) + 1,
HITCBC 1284   1 s_ + impl_.offset(id_path), n); 1294   1 s_ + impl_.offset(id_path), n);
HITCBC 1285   1 *(s_ + impl_.offset(id_path)) = '.'; 1295   1 *(s_ + impl_.offset(id_path)) = '.';
HITCBC 1286   1 ++impl_.decoded_[id_path]; 1296   1 ++impl_.decoded_[id_path];
HITCBC 1287   1 return true; 1297   1 return true;
1288   } 1298   }
1289   1299  
1290   // remove '/' 1300   // remove '/'
HITCBC 1291   4 auto n = impl_.len(id_port); 1301   4 auto n = impl_.len(id_port);
HITCBC 1292   4 impl_.split(id_port, n + 1); 1302   4 impl_.split(id_port, n + 1);
HITCBC 1293   4 resize_impl(id_port, n, op); 1303   4 resize_impl(id_port, n, op);
HITCBC 1294   4 --impl_.decoded_[id_path]; 1304   4 --impl_.decoded_[id_path];
HITCBC 1295   4 return true; 1305   4 return true;
1296   } 1306   }
1297   1307  
HITCBC 1298   3 if(! absolute) 1308   3 if(! absolute)
1299   { 1309   {
1300   // already not absolute 1310   // already not absolute
HITCBC 1301   1 return true; 1311   1 return true;
1302   } 1312   }
1303   1313  
1304   // add '/' 1314   // add '/'
HITCBC 1305   2 auto n = impl_.len(id_port); 1315   2 auto n = impl_.len(id_port);
HITCBC 1306   2 auto dest = resize_impl( 1316   2 auto dest = resize_impl(
HITCBC 1307   2 id_port, n + 1, op) + n; 1317   2 id_port, n + 1, op) + n;
HITCBC 1308   2 impl_.split(id_port, n); 1318   2 impl_.split(id_port, n);
HITCBC 1309   2 *dest = '/'; 1319   2 *dest = '/';
HITCBC 1310   2 ++impl_.decoded_[id_path]; 1320   2 ++impl_.decoded_[id_path];
HITCBC 1311   2 return true; 1321   2 return true;
HITCBC 1312   52 } 1322   52 }
1313   1323  
1314   inline 1324   inline
1315   url_base& 1325   url_base&
HITCBC 1316   298 url_base:: 1326   298 url_base::
1317   set_path( 1327   set_path(
1318   core::string_view s) 1328   core::string_view s)
1319   { 1329   {
HITCBC 1320   298 op_t op(*this, &s); 1330   298 op_t op(*this, &s);
HITCBC 1321   298 encoding_opts opt; 1331   298 encoding_opts opt;
1322   1332  
1323   //------------------------------------------------ 1333   //------------------------------------------------
1324   // 1334   //
1325   // Calculate encoded size 1335   // Calculate encoded size
1326   // 1336   //
1327   // - "/"s are not encoded 1337   // - "/"s are not encoded
1328   // - "%2F"s are not encoded 1338   // - "%2F"s are not encoded
1329   // 1339   //
1330   // - reserved path chars are re-encoded 1340   // - reserved path chars are re-encoded
1331   // - colons in first segment might need to be re-encoded 1341   // - colons in first segment might need to be re-encoded
1332   // - the path might need to receive a prefix 1342   // - the path might need to receive a prefix
HITCBC 1333   298 auto const n = encoded_size( 1343   298 auto const n = encoded_size(
1334   s, detail::path_chars, opt); 1344   s, detail::path_chars, opt);
HITCBC 1335   298 std::size_t n_reencode_colons = 0; 1345   298 std::size_t n_reencode_colons = 0;
HITCBC 1336   298 core::string_view first_seg; 1346   298 core::string_view first_seg;
HITCBC 1337   298 if (!has_scheme() && 1347   298 if (!has_scheme() &&
HITCBC 1338   337 !has_authority() && 1348   337 !has_authority() &&
HITCBC 1339   39 !s.starts_with('/')) 1349   39 !s.starts_with('/'))
1340   { 1350   {
1341   // the first segment with unencoded colons would look 1351   // the first segment with unencoded colons would look
1342   // like the scheme 1352   // like the scheme
HITCBC 1343   6 first_seg = detail::to_sv(s); 1353   6 first_seg = detail::to_sv(s);
HITCBC 1344   6 std::size_t p = s.find('/'); 1354   6 std::size_t p = s.find('/');
HITCBC 1345   6 if (p != core::string_view::npos) 1355   6 if (p != core::string_view::npos)
HITCBC 1346   2 first_seg = s.substr(0, p); 1356   2 first_seg = s.substr(0, p);
HITCBC 1347   6 n_reencode_colons = std::count( 1357   6 n_reencode_colons = std::count(
HITCBC 1348   12 first_seg.begin(), first_seg.end(), ':'); 1358   12 first_seg.begin(), first_seg.end(), ':');
1349   } 1359   }
1350   // the authority can only be followed by an empty or relative path 1360   // the authority can only be followed by an empty or relative path
1351   // if we have an authority and the path is a non-empty relative path, we 1361   // if we have an authority and the path is a non-empty relative path, we
1352   // add the "/" prefix to make it valid. 1362   // add the "/" prefix to make it valid.
1353   bool make_absolute = 1363   bool make_absolute =
HITCBC 1354   298 has_authority() && 1364   298 has_authority() &&
HITCBC 1355   305 !s.starts_with('/') && 1365   305 !s.starts_with('/') &&
HITCBC 1356   7 !s.empty(); 1366   7 !s.empty();
1357   // a path starting with "//" might look like the authority. 1367   // a path starting with "//" might look like the authority.
1358   // we add a "/." prefix to prevent that 1368   // we add a "/." prefix to prevent that
1359   bool add_dot_segment = 1369   bool add_dot_segment =
HITCBC 1360   592 !make_absolute && 1370   592 !make_absolute &&
HITCBC 1361   294 s.starts_with("//"); 1371   294 s.starts_with("//");
1362   1372  
1363   //------------------------------------------------ 1373   //------------------------------------------------
1364   // 1374   //
1365   // Re-encode data 1375   // Re-encode data
1366   // 1376   //
HITCBC 1367   596 auto dest = set_path_impl( 1377   596 auto dest = set_path_impl(
HITCBC 1368   298 n + make_absolute + 2 * n_reencode_colons + 2 * add_dot_segment, op); 1378   298 n + make_absolute + 2 * n_reencode_colons + 2 * add_dot_segment, op);
HITCBC 1369   298 impl_.decoded_[id_path] = 0; 1379   298 impl_.decoded_[id_path] = 0;
HITCBC 1370   298 if (!dest) 1380   298 if (!dest)
1371   { 1381   {
HITCBC 1372   3 impl_.nseg_ = 0; 1382   3 impl_.nseg_ = 0;
HITCBC 1373   3 return *this; 1383   3 return *this;
1374   } 1384   }
HITCBC 1375   295 if (make_absolute) 1385   295 if (make_absolute)
1376   { 1386   {
HITCBC 1377   4 *dest++ = '/'; 1387   4 *dest++ = '/';
HITCBC 1378   4 impl_.decoded_[id_path] += 1; 1388   4 impl_.decoded_[id_path] += 1;
1379   } 1389   }
HITCBC 1380   291 else if (add_dot_segment) 1390   291 else if (add_dot_segment)
1381   { 1391   {
HITCBC 1382   12 *dest++ = '/'; 1392   12 *dest++ = '/';
HITCBC 1383   12 *dest++ = '.'; 1393   12 *dest++ = '.';
HITCBC 1384   12 impl_.decoded_[id_path] += 2; 1394   12 impl_.decoded_[id_path] += 2;
1385   } 1395   }
HITCBC 1386   590 dest += encode_unsafe( 1396   590 dest += encode_unsafe(
1387   dest, 1397   dest,
HITCBC 1388   295 impl_.get(id_query).data() - dest, 1398   295 impl_.get(id_query).data() - dest,
1389   first_seg, 1399   first_seg,
HITCBC 1390   295 detail::segment_chars - ':', 1400   295 detail::segment_chars - ':',
1391   opt); 1401   opt);
HITCBC 1392   295 dest += encode_unsafe( 1402   295 dest += encode_unsafe(
1393   dest, 1403   dest,
HITCBC 1394   295 impl_.get(id_query).data() - dest, 1404   295 impl_.get(id_query).data() - dest,
1395   s.substr(first_seg.size()), 1405   s.substr(first_seg.size()),
1396   detail::path_chars, 1406   detail::path_chars,
1397   opt); 1407   opt);
HITCBC 1398   295 impl_.decoded_[id_path] += 1408   295 impl_.decoded_[id_path] +=
HITCBC 1399   295 detail::to_size_type(s.size()); 1409   295 detail::to_size_type(s.size());
HITCBC 1400   295 BOOST_ASSERT(!dest || dest == impl_.get(id_query).data()); 1410   295 BOOST_ASSERT(!dest || dest == impl_.get(id_query).data());
HITCBC 1401   295 BOOST_ASSERT( 1411   295 BOOST_ASSERT(
1402   impl_.decoded_[id_path] == 1412   impl_.decoded_[id_path] ==
1403   s.size() + make_absolute + 2 * add_dot_segment); 1413   s.size() + make_absolute + 2 * add_dot_segment);
1404   1414  
1405   //------------------------------------------------ 1415   //------------------------------------------------
1406   // 1416   //
1407   // Update path parameters 1417   // Update path parameters
1408   // 1418   //
1409   // get the encoded_path with the replacements we applied 1419   // get the encoded_path with the replacements we applied
HITCBC 1410   295 if (s == "/") 1420   295 if (s == "/")
1411   { 1421   {
1412   // "/" maps to sequence {} 1422   // "/" maps to sequence {}
HITCBC 1413   144 impl_.nseg_ = 0; 1423   144 impl_.nseg_ = 0;
1414   } 1424   }
HITCBC 1415   151 else if (!s.empty()) 1425   151 else if (!s.empty())
1416   { 1426   {
HITCBC 1417   146 if (s.starts_with("/./")) 1427   146 if (s.starts_with("/./"))
HITCBC 1418   1 s = s.substr(2); 1428   1 s = s.substr(2);
1419   // count segments as number of '/'s + 1 1429   // count segments as number of '/'s + 1
HITCBC 1420   292 impl_.nseg_ = detail::to_size_type( 1430   292 impl_.nseg_ = detail::to_size_type(
HITCBC 1421   146 std::count( 1431   146 std::count(
HITCBC 1422   292 s.begin() + 1, s.end(), '/') + 1); 1432   292 s.begin() + 1, s.end(), '/') + 1);
1423   } 1433   }
1424   else 1434   else
1425   { 1435   {
1426   // an empty relative path maps to sequence {} 1436   // an empty relative path maps to sequence {}
HITCBC 1427   5 impl_.nseg_ = 0; 1437   5 impl_.nseg_ = 0;
1428   } 1438   }
1429   1439  
HITCBC 1430   295 check_invariants(); 1440   295 check_invariants();
HITCBC 1431   295 return *this; 1441   295 return *this;
HITCBC 1432   298 } 1442   298 }
1433   1443  
1434   inline 1444   inline
1435   url_base& 1445   url_base&
HITCBC 1436   299 url_base:: 1446   300 url_base::
1437   set_encoded_path( 1447   set_encoded_path(
1438   pct_string_view s) 1448   pct_string_view s)
1439   { 1449   {
HITCBC 1440   299 op_t op(*this, &detail::ref(s)); 1450   300 op_t op(*this, &detail::ref(s));
1441   1451  
1442   //------------------------------------------------ 1452   //------------------------------------------------
1443   // 1453   //
1444   // Calculate re-encoded output size 1454   // Calculate re-encoded output size
1445   // 1455   //
1446   // - reserved path chars are re-encoded 1456   // - reserved path chars are re-encoded
1447   // - colons in first segment might need to be re-encoded 1457   // - colons in first segment might need to be re-encoded
1448   // - the path might need to receive a prefix 1458   // - the path might need to receive a prefix
HITCBC 1449   299 auto const n = detail::re_encoded_size_unsafe( 1459   300 auto const n = detail::re_encoded_size_unsafe(
1450   s, detail::path_chars); 1460   s, detail::path_chars);
HITCBC 1451   299 std::size_t n_reencode_colons = 0; 1461   300 std::size_t n_reencode_colons = 0;
HITCBC 1452   299 core::string_view first_seg; 1462   300 core::string_view first_seg;
HITCBC 1453   299 if (!has_scheme() && 1463   300 if (!has_scheme() &&
HITCBC 1454   332 !has_authority() && 1464   333 !has_authority() &&
HITCBC 1455   33 !s.starts_with('/')) 1465   33 !s.starts_with('/'))
1456   { 1466   {
1457   // the first segment with unencoded colons would look 1467   // the first segment with unencoded colons would look
1458   // like the scheme 1468   // like the scheme
HITCBC 1459   26 first_seg = detail::to_sv(s); 1469   26 first_seg = detail::to_sv(s);
HITCBC 1460   26 std::size_t p = s.find('/'); 1470   26 std::size_t p = s.find('/');
HITCBC 1461   26 if (p != core::string_view::npos) 1471   26 if (p != core::string_view::npos)
HITCBC 1462   9 first_seg = s.substr(0, p); 1472   9 first_seg = s.substr(0, p);
HITCBC 1463   26 n_reencode_colons = std::count( 1473   26 n_reencode_colons = std::count(
HITCBC 1464   52 first_seg.begin(), first_seg.end(), ':'); 1474   52 first_seg.begin(), first_seg.end(), ':');
1465   } 1475   }
1466   // the authority can only be followed by an empty or relative path 1476   // the authority can only be followed by an empty or relative path
1467   // if we have an authority and the path is a non-empty relative path, we 1477   // if we have an authority and the path is a non-empty relative path, we
1468   // add the "/" prefix to make it valid. 1478   // add the "/" prefix to make it valid.
1469   bool make_absolute = 1479   bool make_absolute =
HITCBC 1470   299 has_authority() && 1480   300 has_authority() &&
HITCBC 1471   443 !s.starts_with('/') && 1481   444 !s.starts_with('/') &&
HITCBC 1472   144 !s.empty(); 1482   144 !s.empty();
1473   // a path starting with "//" might look like the authority 1483   // a path starting with "//" might look like the authority
1474   // we add a "/." prefix to prevent that 1484   // we add a "/." prefix to prevent that
1475   bool add_dot_segment = 1485   bool add_dot_segment =
HITCBC 1476   502 !make_absolute && 1486   504 !make_absolute &&
HITCBC 1477   362 !has_authority() && 1487   363 !has_authority() &&
HITCBC 1478   63 s.starts_with("//"); 1488   63 s.starts_with("//");
1479   1489  
1480   //------------------------------------------------ 1490   //------------------------------------------------
1481   // 1491   //
1482   // Re-encode data 1492   // Re-encode data
1483   // 1493   //
HITCBC 1484   598 auto dest = set_path_impl( 1494   600 auto dest = set_path_impl(
HITCBC 1485   299 n + make_absolute + 2 * n_reencode_colons + 2 * add_dot_segment, op); 1495   300 n + make_absolute + 2 * n_reencode_colons + 2 * add_dot_segment, op);
HITCBC 1486   299 impl_.decoded_[id_path] = 0; 1496   300 impl_.decoded_[id_path] = 0;
HITCBC 1487   299 if (!dest) 1497   300 if (!dest)
1488   { 1498   {
HITCBC 1489   2 impl_.nseg_ = 0; 1499   2 impl_.nseg_ = 0;
HITCBC 1490   2 return *this; 1500   2 return *this;
1491   } 1501   }
HITCBC 1492   297 if (make_absolute) 1502   298 if (make_absolute)
1493   { 1503   {
HITCBC 1494   96 *dest++ = '/'; 1504   96 *dest++ = '/';
HITCBC 1495   96 impl_.decoded_[id_path] += 1; 1505   96 impl_.decoded_[id_path] += 1;
1496   } 1506   }
HITCBC 1497   201 else if (add_dot_segment) 1507   202 else if (add_dot_segment)
1498   { 1508   {
HITCBC 1499   5 *dest++ = '/'; 1509   5 *dest++ = '/';
HITCBC 1500   5 *dest++ = '.'; 1510   5 *dest++ = '.';
HITCBC 1501   5 impl_.decoded_[id_path] += 2; 1511   5 impl_.decoded_[id_path] += 2;
1502   } 1512   }
HITCBC 1503   297 impl_.decoded_[id_path] += 1513   298 impl_.decoded_[id_path] +=
HITCBC 1504   297 detail::to_size_type(detail::re_encode_unsafe( 1514   298 detail::to_size_type(detail::re_encode_unsafe(
1505   dest, 1515   dest,
HITCBC 1506   297 impl_.get(id_query).data(), 1516   298 impl_.get(id_query).data(),
1507   first_seg, 1517   first_seg,
HITCBC 1508   297 detail::segment_chars - ':')); 1518   298 detail::segment_chars - ':'));
HITCBC 1509   297 impl_.decoded_[id_path] += 1519   298 impl_.decoded_[id_path] +=
HITCBC 1510   594 detail::to_size_type(detail::re_encode_unsafe( 1520   596 detail::to_size_type(detail::re_encode_unsafe(
1511   dest, 1521   dest,
HITCBC 1512   297 impl_.get(id_query).data(), 1522   298 impl_.get(id_query).data(),
1513   s.substr(first_seg.size()), 1523   s.substr(first_seg.size()),
1514   detail::path_chars)); 1524   detail::path_chars));
HITCBC 1515   297 BOOST_ASSERT(dest == impl_.get(id_query).data()); 1525   298 BOOST_ASSERT(dest == impl_.get(id_query).data());
HITCBC 1516   297 BOOST_ASSERT( 1526   298 BOOST_ASSERT(
1517   impl_.decoded_[id_path] == 1527   impl_.decoded_[id_path] ==
1518   s.decoded_size() + make_absolute + 2 * add_dot_segment); 1528   s.decoded_size() + make_absolute + 2 * add_dot_segment);
1519   1529  
1520   //------------------------------------------------ 1530   //------------------------------------------------
1521   // 1531   //
1522   // Update path parameters 1532   // Update path parameters
1523   // 1533   //
1524   // get the encoded_path with the replacements we applied 1534   // get the encoded_path with the replacements we applied
HITCBC 1525   297 if (s == "/") 1535   298 if (s == "/")
1526   { 1536   {
1527   // "/" maps to sequence {} 1537   // "/" maps to sequence {}
HITCBC 1528   19 impl_.nseg_ = 0; 1538   20 impl_.nseg_ = 0;
1529   } 1539   }
HITCBC 1530   278 else if (!s.empty()) 1540   278 else if (!s.empty())
1531   { 1541   {
HITCBC 1532   227 if (s.starts_with("/./")) 1542   227 if (s.starts_with("/./"))
HITCBC 1533   7 s = s.substr(2); 1543   7 s = s.substr(2);
1534   // count segments as number of '/'s + 1 1544   // count segments as number of '/'s + 1
HITCBC 1535   454 impl_.nseg_ = detail::to_size_type( 1545   454 impl_.nseg_ = detail::to_size_type(
HITCBC 1536   227 std::count( 1546   227 std::count(
HITCBC 1537   454 s.begin() + 1, s.end(), '/') + 1); 1547   454 s.begin() + 1, s.end(), '/') + 1);
1538   } 1548   }
1539   else 1549   else
1540   { 1550   {
1541   // an empty relative path maps to sequence {} 1551   // an empty relative path maps to sequence {}
HITCBC 1542   51 impl_.nseg_ = 0; 1552   51 impl_.nseg_ = 0;
1543   } 1553   }
1544   1554  
HITCBC 1545   297 check_invariants(); 1555   298 check_invariants();
HITCBC 1546   297 return *this; 1556   298 return *this;
HITCBC 1547   299 } 1557   300 }
1548   1558  
1549   inline 1559   inline
1550   segments_ref 1560   segments_ref
HITCBC 1551   1146 url_base:: 1561   1146 url_base::
1552   segments() noexcept 1562   segments() noexcept
1553   { 1563   {
HITCBC 1554   1146 return {*this}; 1564   1146 return {*this};
1555   } 1565   }
1556   1566  
1557   inline 1567   inline
1558   segments_encoded_ref 1568   segments_encoded_ref
HITCBC 1559   497 url_base:: 1569   522 url_base::
1560   encoded_segments() noexcept 1570   encoded_segments() noexcept
1561   { 1571   {
HITCBC 1562   497 return {*this}; 1572   522 return {*this};
1563   } 1573   }
1564   1574  
1565   //------------------------------------------------ 1575   //------------------------------------------------
1566   // 1576   //
1567   // Query 1577   // Query
1568   // 1578   //
1569   //------------------------------------------------ 1579   //------------------------------------------------
1570   1580  
1571   inline 1581   inline
1572   url_base& 1582   url_base&
HITCBC 1573   169 url_base:: 1583   169 url_base::
1574   set_query( 1584   set_query(
1575   core::string_view s) 1585   core::string_view s)
1576   { 1586   {
HITCBC 1577   169 edit_params( 1587   169 edit_params(
HITCBC 1578   169 detail::params_iter_impl(impl_), 1588   169 detail::params_iter_impl(impl_),
HITCBC 1579   169 detail::params_iter_impl(impl_, 0), 1589   169 detail::params_iter_impl(impl_, 0),
HITCBC 1580   338 detail::query_string_iter(s, true)); 1590   338 detail::query_string_iter(s, true));
HITCBC 1581   169 return *this; 1591   169 return *this;
1582   } 1592   }
1583   1593  
1584   inline 1594   inline
1585   url_base& 1595   url_base&
HITCBC 1586   179 url_base:: 1596   179 url_base::
1587   set_encoded_query( 1597   set_encoded_query(
1588   pct_string_view s) 1598   pct_string_view s)
1589   { 1599   {
HITCBC 1590   179 op_t op(*this); 1600   179 op_t op(*this);
HITCBC 1591   179 std::size_t n = 0; // encoded size 1601   179 std::size_t n = 0; // encoded size
HITCBC 1592   179 std::size_t nparam = 1; // param count 1602   179 std::size_t nparam = 1; // param count
HITCBC 1593   179 auto const end = s.end(); 1603   179 auto const end = s.end();
HITCBC 1594   179 auto p = s.begin(); 1604   179 auto p = s.begin();
1595   1605  
1596   // measure 1606   // measure
HITCBC 1597   742 while(p != end) 1607   742 while(p != end)
1598   { 1608   {
HITCBC 1599   563 if(*p == '&') 1609   563 if(*p == '&')
1600   { 1610   {
HITCBC 1601   4 ++p; 1611   4 ++p;
HITCBC 1602   4 ++n; 1612   4 ++n;
HITCBC 1603   4 ++nparam; 1613   4 ++nparam;
1604   } 1614   }
HITCBC 1605   559 else if(*p != '%') 1615   559 else if(*p != '%')
1606   { 1616   {
HITCBC 1607   533 if(detail::query_chars(*p)) 1617   533 if(detail::query_chars(*p))
HITCBC 1608   518 n += 1; // allowed 1618   518 n += 1; // allowed
1609   else 1619   else
HITCBC 1610   15 n += 3; // escaped 1620   15 n += 3; // escaped
HITCBC 1611   533 ++p; 1621   533 ++p;
1612   } 1622   }
1613   else 1623   else
1614   { 1624   {
1615   // escape 1625   // escape
HITCBC 1616   26 n += 3; 1626   26 n += 3;
HITCBC 1617   26 p += 3; 1627   26 p += 3;
1618   } 1628   }
1619   } 1629   }
1620   1630  
1621   // resize 1631   // resize
HITCBC 1622   179 auto dest = resize_impl( 1632   179 auto dest = resize_impl(
HITCBC 1623   179 id_query, n + 1, op); 1633   179 id_query, n + 1, op);
HITCBC 1624   179 *dest++ = '?'; 1634   179 *dest++ = '?';
1625   1635  
1626   // encode 1636   // encode
HITCBC 1627   179 impl_.decoded_[id_query] = 1637   179 impl_.decoded_[id_query] =
HITCBC 1628   358 detail::to_size_type(detail::re_encode_unsafe( 1638   358 detail::to_size_type(detail::re_encode_unsafe(
1629   dest, 1639   dest,
HITCBC 1630   179 dest + n, 1640   179 dest + n,
1631   s, 1641   s,
1632   detail::query_chars)); 1642   detail::query_chars));
HITCBC 1633   179 BOOST_ASSERT( 1643   179 BOOST_ASSERT(
1634   impl_.decoded_[id_query] == 1644   impl_.decoded_[id_query] ==
1635   s.decoded_size()); 1645   s.decoded_size());
HITCBC 1636   179 impl_.nparam_ = 1646   179 impl_.nparam_ =
HITCBC 1637   179 detail::to_size_type(nparam); 1647   179 detail::to_size_type(nparam);
HITCBC 1638   179 return *this; 1648   179 return *this;
HITCBC 1639   179 } 1649   179 }
1640   1650  
1641   inline 1651   inline
1642   params_ref 1652   params_ref
HITCBC 1643   986 url_base:: 1653   986 url_base::
1644   params() noexcept 1654   params() noexcept
1645   { 1655   {
1646   return params_ref( 1656   return params_ref(
1647   *this, 1657   *this,
1648   encoding_opts{ 1658   encoding_opts{
HITCBC 1649   986 true, false, false}); 1659   986 true, false, false});
1650   } 1660   }
1651   1661  
1652   inline 1662   inline
1653   params_ref 1663   params_ref
HITCBC 1654   4 url_base:: 1664   4 url_base::
1655   params(encoding_opts opt) noexcept 1665   params(encoding_opts opt) noexcept
1656   { 1666   {
HITCBC 1657   4 return {*this, opt}; 1667   4 return {*this, opt};
1658   } 1668   }
1659   1669  
1660   inline 1670   inline
1661   params_encoded_ref 1671   params_encoded_ref
HITCBC 1662   77 url_base:: 1672   77 url_base::
1663   encoded_params() noexcept 1673   encoded_params() noexcept
1664   { 1674   {
HITCBC 1665   77 return {*this}; 1675   77 return {*this};
1666   } 1676   }
1667   1677  
1668   inline 1678   inline
1669   url_base& 1679   url_base&
HITCBC 1670   1 url_base:: 1680   1 url_base::
1671   set_params( 1681   set_params(
1672   std::initializer_list<param_view> ps, 1682   std::initializer_list<param_view> ps,
1673   encoding_opts opts) noexcept 1683   encoding_opts opts) noexcept
1674   { 1684   {
HITCBC 1675   1 params(opts).assign(ps); 1685   1 params(opts).assign(ps);
HITCBC 1676   1 return *this; 1686   1 return *this;
1677   } 1687   }
1678   1688  
1679   inline 1689   inline
1680   url_base& 1690   url_base&
HITCBC 1681   1 url_base:: 1691   1 url_base::
1682   set_encoded_params( std::initializer_list< param_pct_view > ps ) noexcept 1692   set_encoded_params( std::initializer_list< param_pct_view > ps ) noexcept
1683   { 1693   {
HITCBC 1684   1 encoded_params().assign(ps); 1694   1 encoded_params().assign(ps);
HITCBC 1685   1 return *this; 1695   1 return *this;
1686   } 1696   }
1687   1697  
1688   inline 1698   inline
1689   url_base& 1699   url_base&
HITCBC 1690   990 url_base:: 1700   991 url_base::
1691   remove_query() noexcept 1701   remove_query() noexcept
1692   { 1702   {
HITCBC 1693   990 op_t op(*this); 1703   991 op_t op(*this);
HITCBC 1694   990 resize_impl(id_query, 0, op); 1704   991 resize_impl(id_query, 0, op);
HITCBC 1695   990 impl_.nparam_ = 0; 1705   991 impl_.nparam_ = 0;
HITCBC 1696   990 impl_.decoded_[id_query] = 0; 1706   991 impl_.decoded_[id_query] = 0;
HITCBC 1697   1980 return *this; 1707   1982 return *this;
HITCBC 1698   990 } 1708   991 }
1699   1709  
1700   //------------------------------------------------ 1710   //------------------------------------------------
1701   // 1711   //
1702   // Fragment 1712   // Fragment
1703   // 1713   //
1704   //------------------------------------------------ 1714   //------------------------------------------------
1705   1715  
1706   inline 1716   inline
1707   url_base& 1717   url_base&
HITCBC 1708   407 url_base:: 1718   408 url_base::
1709   remove_fragment() noexcept 1719   remove_fragment() noexcept
1710   { 1720   {
HITCBC 1711   407 op_t op(*this); 1721   408 op_t op(*this);
HITCBC 1712   407 resize_impl(id_frag, 0, op); 1722   408 resize_impl(id_frag, 0, op);
HITCBC 1713   407 impl_.decoded_[id_frag] = 0; 1723   408 impl_.decoded_[id_frag] = 0;
HITCBC 1714   814 return *this; 1724   816 return *this;
HITCBC 1715   407 } 1725   408 }
1716   1726  
1717   inline 1727   inline
1718   url_base& 1728   url_base&
HITCBC 1719   151 url_base:: 1729   151 url_base::
1720   set_fragment(core::string_view s) 1730   set_fragment(core::string_view s)
1721   { 1731   {
HITCBC 1722   151 op_t op(*this, &s); 1732   151 op_t op(*this, &s);
HITCBC 1723   151 encoding_opts opt; 1733   151 encoding_opts opt;
HITCBC 1724   151 auto const n = encoded_size( 1734   151 auto const n = encoded_size(
1725   s, 1735   s,
1726   detail::fragment_chars, 1736   detail::fragment_chars,
1727   opt); 1737   opt);
HITCBC 1728   151 auto dest = resize_impl( 1738   151 auto dest = resize_impl(
1729   id_frag, n + 1, op); 1739   id_frag, n + 1, op);
HITCBC 1730   151 *dest++ = '#'; 1740   151 *dest++ = '#';
HITCBC 1731   151 encode_unsafe( 1741   151 encode_unsafe(
1732   dest, 1742   dest,
1733   n, 1743   n,
1734   s, 1744   s,
1735   detail::fragment_chars, 1745   detail::fragment_chars,
1736   opt); 1746   opt);
HITCBC 1737   151 impl_.decoded_[id_frag] = 1747   151 impl_.decoded_[id_frag] =
HITCBC 1738   151 detail::to_size_type(s.size()); 1748   151 detail::to_size_type(s.size());
HITCBC 1739   151 return *this; 1749   151 return *this;
HITCBC 1740   151 } 1750   151 }
1741   1751  
1742   inline 1752   inline
1743   url_base& 1753   url_base&
HITCBC 1744   183 url_base:: 1754   183 url_base::
1745   set_encoded_fragment( 1755   set_encoded_fragment(
1746   pct_string_view s) 1756   pct_string_view s)
1747   { 1757   {
HITCBC 1748   183 op_t op(*this, &detail::ref(s)); 1758   183 op_t op(*this, &detail::ref(s));
1749   auto const n = 1759   auto const n =
HITCBC 1750   183 detail::re_encoded_size_unsafe( 1760   183 detail::re_encoded_size_unsafe(
1751   s, 1761   s,
1752   detail::fragment_chars); 1762   detail::fragment_chars);
HITCBC 1753   183 auto dest = resize_impl( 1763   183 auto dest = resize_impl(
HITCBC 1754   183 id_frag, n + 1, op); 1764   183 id_frag, n + 1, op);
HITCBC 1755   183 *dest++ = '#'; 1765   183 *dest++ = '#';
HITCBC 1756   183 impl_.decoded_[id_frag] = 1766   183 impl_.decoded_[id_frag] =
HITCBC 1757   366 detail::to_size_type(detail::re_encode_unsafe( 1767   366 detail::to_size_type(detail::re_encode_unsafe(
1758   dest, 1768   dest,
HITCBC 1759   183 dest + n, 1769   183 dest + n,
1760   s, 1770   s,
1761   detail::fragment_chars)); 1771   detail::fragment_chars));
HITCBC 1762   183 BOOST_ASSERT( 1772   183 BOOST_ASSERT(
1763   impl_.decoded_[id_frag] == 1773   impl_.decoded_[id_frag] ==
1764   s.decoded_size()); 1774   s.decoded_size());
HITCBC 1765   183 return *this; 1775   183 return *this;
HITCBC 1766   183 } 1776   183 }
1767   1777  
1768   //------------------------------------------------ 1778   //------------------------------------------------
1769   // 1779   //
1770   // Resolution 1780   // Resolution
1771   // 1781   //
1772   //------------------------------------------------ 1782   //------------------------------------------------
1773   1783  
1774   inline 1784   inline
1775   system::result<void> 1785   system::result<void>
HITCBC 1776   516 url_base:: 1786   517 url_base::
1777   resolve( 1787   resolve(
1778   url_view_base const& ref) 1788   url_view_base const& ref)
1779   { 1789   {
HITCBC 1780   516 if(! has_scheme()) 1790   517 if(! has_scheme())
1781   { 1791   {
HITCBC 1782   2 BOOST_URL_RETURN_EC(error::not_a_base); 1792   2 BOOST_URL_RETURN_EC(error::not_a_base);
1783   } 1793   }
1784   1794  
HITCBC 1785   514 if (this == &ref) 1795   515 if (this == &ref)
1786   { 1796   {
HITCBC 1787   2 normalize_path(); 1797   2 normalize_path();
HITCBC 1788   2 return {}; 1798   2 return {};
1789   } 1799   }
1790   1800  
HITCBC 1791   512 op_t op(*this); 1801   513 op_t op(*this);
1792   1802  
1793   // 1803   //
1794   // 5.2.2. Transform References 1804   // 5.2.2. Transform References
1795   // https://datatracker.ietf.org/doc/html/rfc3986#section-5.2.2 1805   // https://datatracker.ietf.org/doc/html/rfc3986#section-5.2.2
1796   // 1806   //
1797   1807  
HITCBC 1798   782 if( ref.has_scheme() && 1808   783 if( ref.has_scheme() &&
HITCBC 1799   270 ref.scheme() != scheme()) 1809   270 ref.scheme() != scheme())
1800   { 1810   {
HITCBC 1801   202 reserve_impl(ref.size(), op); 1811   202 reserve_impl(ref.size(), op);
HITCBC 1802   202 copy(ref); 1812   202 copy(ref);
HITCBC 1803   202 normalize_path(); 1813   202 normalize_path();
HITCBC 1804   202 return {}; 1814   202 return {};
1805   } 1815   }
HITCBC 1806   310 if(ref.has_authority()) 1816   311 if(ref.has_authority())
1807   { 1817   {
HITCBC 1808   78 reserve_impl( 1818   79 reserve_impl(
HITCBC 1809   78 impl_.offset(id_user) + ref.size(), op); 1819   79 impl_.offset(id_user) + ref.size(), op);
HITCBC 1810   78 set_encoded_authority( 1820   79 set_encoded_authority(
1811   ref.encoded_authority()); 1821   ref.encoded_authority());
HITCBC 1812   78 set_encoded_path( 1822   79 set_encoded_path(
1813   ref.encoded_path()); 1823   ref.encoded_path());
HITCBC 1814   78 if (ref.encoded_path().empty()) 1824   79 if (ref.encoded_path().empty())
HITCBC 1815   33 set_path_absolute(false); 1825   33 set_path_absolute(false);
1816   else 1826   else
HITCBC 1817   45 normalize_path(); 1827   46 normalize_path();
HITCBC 1818   78 if(ref.has_query()) 1828   79 if(ref.has_query())
HITCBC 1819   8 set_encoded_query( 1829   8 set_encoded_query(
1820   ref.encoded_query()); 1830   ref.encoded_query());
1821   else 1831   else
HITCBC 1822   70 remove_query(); 1832   71 remove_query();
HITCBC 1823   78 if(ref.has_fragment()) 1833   79 if(ref.has_fragment())
HITCBC 1824   7 set_encoded_fragment( 1834   7 set_encoded_fragment(
1825   ref.encoded_fragment()); 1835   ref.encoded_fragment());
1826   else 1836   else
HITCBC 1827   71 remove_fragment(); 1837   72 remove_fragment();
HITCBC 1828   78 return {}; 1838   79 return {};
1829   } 1839   }
HITCBC 1830   232 if(ref.encoded_path().empty()) 1840   232 if(ref.encoded_path().empty())
1831   { 1841   {
HITCBC 1832   43 reserve_impl( 1842   43 reserve_impl(
HITCBC 1833   43 impl_.offset(id_query) + 1843   43 impl_.offset(id_query) +
HITCBC 1834   43 ref.size(), op); 1844   43 ref.size(), op);
HITCBC 1835   43 normalize_path(); 1845   43 normalize_path();
HITCBC 1836   43 if(ref.has_query()) 1846   43 if(ref.has_query())
1837   { 1847   {
HITCBC 1838   12 set_encoded_query( 1848   12 set_encoded_query(
1839   ref.encoded_query()); 1849   ref.encoded_query());
1840   } 1850   }
HITCBC 1841   43 if(ref.has_fragment()) 1851   43 if(ref.has_fragment())
HITCBC 1842   20 set_encoded_fragment( 1852   20 set_encoded_fragment(
1843   ref.encoded_fragment()); 1853   ref.encoded_fragment());
1844   else 1854   else
HITCBC 1845   23 remove_fragment(); 1855   23 remove_fragment();
HITCBC 1846   43 return {}; 1856   43 return {};
1847   } 1857   }
HITCBC 1848   189 if(ref.is_path_absolute()) 1858   189 if(ref.is_path_absolute())
1849   { 1859   {
HITCBC 1850   41 reserve_impl( 1860   41 reserve_impl(
HITCBC 1851   41 impl_.offset(id_path) + 1861   41 impl_.offset(id_path) +
HITCBC 1852   41 ref.size(), op); 1862   41 ref.size(), op);
HITCBC 1853   41 set_encoded_path( 1863   41 set_encoded_path(
1854   ref.encoded_path()); 1864   ref.encoded_path());
HITCBC 1855   41 normalize_path(); 1865   41 normalize_path();
HITCBC 1856   41 if(ref.has_query()) 1866   41 if(ref.has_query())
HITCBC 1857   4 set_encoded_query( 1867   4 set_encoded_query(
1858   ref.encoded_query()); 1868   ref.encoded_query());
1859   else 1869   else
HITCBC 1860   37 remove_query(); 1870   37 remove_query();
HITCBC 1861   41 if(ref.has_fragment()) 1871   41 if(ref.has_fragment())
HITCBC 1862   2 set_encoded_fragment( 1872   2 set_encoded_fragment(
1863   ref.encoded_fragment()); 1873   ref.encoded_fragment());
1864   else 1874   else
HITCBC 1865   39 remove_fragment(); 1875   39 remove_fragment();
HITCBC 1866   41 return {}; 1876   41 return {};
1867   } 1877   }
1868   // General case: ref is relative path 1878   // General case: ref is relative path
HITCBC 1869   148 reserve_impl( 1879   148 reserve_impl(
HITCBC 1870   148 impl_.offset(id_query) + 1880   148 impl_.offset(id_query) +
HITCBC 1871   148 ref.size(), op); 1881   148 ref.size(), op);
1872   // 5.2.3. Merge Paths 1882   // 5.2.3. Merge Paths
HITCBC 1873   148 auto es = encoded_segments(); 1883   148 auto es = encoded_segments();
HITCBC 1874   148 if(es.size() > 0) 1884   148 if(es.size() > 0)
1875   { 1885   {
HITCBC 1876   141 es.pop_back(); 1886   141 es.pop_back();
1877   } 1887   }
HITCBC 1878   296 es.insert(es.end(), 1888   296 es.insert(es.end(),
HITCBC 1879   148 ref.encoded_segments().begin(), 1889   148 ref.encoded_segments().begin(),
HITCBC 1880   148 ref.encoded_segments().end()); 1890   148 ref.encoded_segments().end());
HITCBC 1881   148 normalize_path(); 1891   148 normalize_path();
HITCBC 1882   148 if(ref.has_query()) 1892   148 if(ref.has_query())
HITCBC 1883   10 set_encoded_query( 1893   10 set_encoded_query(
1884   ref.encoded_query()); 1894   ref.encoded_query());
1885   else 1895   else
HITCBC 1886   138 remove_query(); 1896   138 remove_query();
HITCBC 1887   148 if(ref.has_fragment()) 1897   148 if(ref.has_fragment())
HITCBC 1888   13 set_encoded_fragment( 1898   13 set_encoded_fragment(
1889   ref.encoded_fragment()); 1899   ref.encoded_fragment());
1890   else 1900   else
HITCBC 1891   135 remove_fragment(); 1901   135 remove_fragment();
HITCBC 1892   148 return {}; 1902   148 return {};
HITCBC 1893   512 } 1903   513 }
1894   1904  
1895   //------------------------------------------------ 1905   //------------------------------------------------
1896   // 1906   //
1897   // Normalization 1907   // Normalization
1898   // 1908   //
1899   //------------------------------------------------ 1909   //------------------------------------------------
1900   1910  
1901   template < 1911   template <
1902   class AllowedCharset, 1912   class AllowedCharset,
1903   class IgnoredCharset> 1913   class IgnoredCharset>
1904   void 1914   void
HITCBC 1905   2990 url_base:: 1915   3181 url_base::
1906   normalize_octets_impl( 1916   normalize_octets_impl(
1907   int id, 1917   int id,
1908   AllowedCharset const& allowed, 1918   AllowedCharset const& allowed,
1909   IgnoredCharset const& ignored, 1919   IgnoredCharset const& ignored,
1910   op_t& op) noexcept 1920   op_t& op) noexcept
1911   { 1921   {
HITCBC 1912   2990 char* it = s_ + impl_.offset(id); 1922   3181 char* it = s_ + impl_.offset(id);
HITCBC 1913   2990 char* end = s_ + impl_.offset(id + 1); 1923   3181 char* end = s_ + impl_.offset(id + 1);
HITCBC 1914   2990 char d = 0; 1924   3181 char d = 0;
HITCBC 1915   2990 char* dest = it; 1925   3181 char* dest = it;
HITCBC 1916   16466 while (it < end) 1926   16827 while (it < end)
1917   { 1927   {
HITCBC 1918   13476 if (*it != '%') 1928   13646 if (*it != '%')
1919   { 1929   {
HITCBC 1920   13258 *dest = *it; 1930   13428 *dest = *it;
HITCBC 1921   13258 ++it; 1931   13428 ++it;
HITCBC 1922   13258 ++dest; 1932   13428 ++dest;
HITCBC 1923   13258 continue; 1933   13428 continue;
1924   } 1934   }
HITCBC 1925   218 BOOST_ASSERT(end - it >= 3); 1935   218 BOOST_ASSERT(end - it >= 3);
1926   1936  
1927   // decode unreserved octets 1937   // decode unreserved octets
HITCBC 1928   218 d = detail::decode_one(it + 1); 1938   218 d = detail::decode_one(it + 1);
HITCBC 1929   314 if (allowed(d) && 1939   314 if (allowed(d) &&
HITCBC 1930   96 !ignored(d)) 1940   96 !ignored(d))
1931   { 1941   {
HITCBC 1932   87 *dest = d; 1942   87 *dest = d;
HITCBC 1933   87 it += 3; 1943   87 it += 3;
HITCBC 1934   87 ++dest; 1944   87 ++dest;
HITCBC 1935   87 continue; 1945   87 continue;
1936   } 1946   }
1937   1947  
1938   // uppercase percent-encoding triplets 1948   // uppercase percent-encoding triplets
HITCBC 1939   131 *dest++ = '%'; 1949   131 *dest++ = '%';
HITCBC 1940   131 ++it; 1950   131 ++it;
HITCBC 1941   131 *dest++ = grammar::to_upper(*it++); 1951   131 *dest++ = grammar::to_upper(*it++);
HITCBC 1942   131 *dest++ = grammar::to_upper(*it++); 1952   131 *dest++ = grammar::to_upper(*it++);
1943   } 1953   }
HITCBC 1944   2990 if (it != dest) 1954   3181 if (it != dest)
1945   { 1955   {
HITCBC 1946   35 auto diff = it - dest; 1956   35 auto diff = it - dest;
HITCBC 1947   35 auto n = impl_.len(id) - diff; 1957   35 auto n = impl_.len(id) - diff;
HITCBC 1948   35 shrink_impl(id, n, op); 1958   35 shrink_impl(id, n, op);
HITCBC 1949   35 s_[size()] = '\0'; 1959   35 s_[size()] = '\0';
1950   } 1960   }
HITCBC 1951   2990 } 1961   3181 }
1952   1962  
1953   template<class CharSet> 1963   template<class CharSet>
1954   void 1964   void
HITCBC 1955   2769 url_base:: 1965   2922 url_base::
1956   normalize_octets_impl( 1966   normalize_octets_impl(
1957   int idx, 1967   int idx,
1958   CharSet const& allowed, 1968   CharSet const& allowed,
1959   op_t& op) noexcept 1969   op_t& op) noexcept
1960   { 1970   {
HITCBC 1961   2769 return normalize_octets_impl( 1971   2922 return normalize_octets_impl(
HITCBC 1962   2769 idx, allowed, detail::empty_chars, op); 1972   2922 idx, allowed, detail::empty_chars, op);
1963   } 1973   }
1964   1974  
1965   inline 1975   inline
1966   url_base& 1976   url_base&
HITCBC 1967   220 url_base:: 1977   258 url_base::
1968   normalize_scheme() 1978   normalize_scheme()
1969   { 1979   {
HITCBC 1970   220 to_lower_impl(id_scheme); 1980   258 to_lower_impl(id_scheme);
HITCBC 1971   220 return *this; 1981   258 return *this;
1972   } 1982   }
1973   1983  
1974   inline 1984   inline
1975   url_base& 1985   url_base&
HITCBC 1976   566 url_base:: 1986   604 url_base::
1977   normalize_authority() 1987   normalize_authority()
1978   { 1988   {
HITCBC 1979   566 op_t op(*this); 1989   604 op_t op(*this);
1980   1990  
1981   // normalize host 1991   // normalize host
HITCBC 1982   566 if (host_type() == urls::host_type::name) 1992   604 if (host_type() == urls::host_type::name)
1983   { 1993   {
HITCBC 1984   351 normalize_octets_impl( 1994   351 normalize_octets_impl(
1985   id_host, 1995   id_host,
1986   detail::reg_name_chars, op); 1996   detail::reg_name_chars, op);
1987   } 1997   }
HITCBC 1988   566 decoded_to_lower_impl(id_host); 1998   604 decoded_to_lower_impl(id_host);
1989   1999  
1990   // normalize password 2000   // normalize password
HITCBC 1991   566 normalize_octets_impl(id_pass, detail::password_chars, op); 2001   604 normalize_octets_impl(id_pass, detail::password_chars, op);
1992   2002  
1993   // normalize user 2003   // normalize user
HITCBC 1994   566 normalize_octets_impl(id_user, detail::user_chars, op); 2004   604 normalize_octets_impl(id_user, detail::user_chars, op);
HITCBC 1995   1132 return *this; 2005   1208 return *this;
HITCBC 1996   566 } 2006   604 }
1997   2007  
1998   inline 2008   inline
1999   url_base& 2009   url_base&
HITCBC 2000   1068 url_base:: 2010   1107 url_base::
2001   normalize_path() 2011   normalize_path()
2002   { 2012   {
HITCBC 2003   1068 op_t op(*this); 2013   1107 op_t op(*this);
2004   2014  
2005   // 2015   //
2006   // Step 1: Percent-encoding normalization 2016   // Step 1: Percent-encoding normalization
2007   // 2017   //
2008   // Decode percent-encoded characters that are 2018   // Decode percent-encoded characters that are
2009   // valid unencoded in path segments (pchar), 2019   // valid unencoded in path segments (pchar),
2010   // and uppercase remaining hex digits. 2020   // and uppercase remaining hex digits.
2011   // 2021   //
2012   // This can introduce: 2022   // This can introduce:
2013   // - New '.' from %2E (dot is unreserved) 2023   // - New '.' from %2E (dot is unreserved)
2014   // - New ':' from %3A (colon is a pchar) 2024   // - New ':' from %3A (colon is a pchar)
2015   // This cannot introduce: 2025   // This cannot introduce:
2016   // - New '/' (%2F stays encoded, '/' is not 2026   // - New '/' (%2F stays encoded, '/' is not
2017   // a segment character) 2027   // a segment character)
2018   // 2028   //
2019   // These new '.' and ':' characters can create 2029   // These new '.' and ':' characters can create
2020   // ambiguities that Steps 2 and 3 must handle. 2030   // ambiguities that Steps 2 and 3 must handle.
2021   // 2031   //
HITCBC 2022   1068 normalize_octets_impl(id_path, detail::segment_chars, op); 2032   1107 normalize_octets_impl(id_path, detail::segment_chars, op);
HITCBC 2023   1068 core::string_view p = impl_.get(id_path); 2033   1107 core::string_view p = impl_.get(id_path);
HITCBC 2024   1068 char* p_dest = s_ + impl_.offset(id_path); 2034   1107 char* p_dest = s_ + impl_.offset(id_path);
HITCBC 2025   1068 char* p_end = s_ + impl_.offset(id_path + 1); 2035   1107 char* p_end = s_ + impl_.offset(id_path + 1);
HITCBC 2026   1068 auto pn = p.size(); 2036   1107 auto pn = p.size();
2027   2037  
2028   // 2038   //
2029   // Step 2: Preserve existing path shields 2039   // Step 2: Preserve existing path shields
2030   // 2040   //
2031   // If the URL has no authority, a path starting 2041   // If the URL has no authority, a path starting
2032   // with "//" would be misinterpreted as an 2042   // with "//" would be misinterpreted as an
2033   // authority when serialized and re-parsed. 2043   // authority when serialized and re-parsed.
2034   // Some paths already have a dot prefix ("/." 2044   // Some paths already have a dot prefix ("/."
2035   // or "./") that shields against this. We 2045   // or "./") that shields against this. We
2036   // preserve them so remove_dot_segments (Step 3) 2046   // preserve them so remove_dot_segments (Step 3)
2037   // does not strip them. Step 1 can also reveal 2047   // does not strip them. Step 1 can also reveal
2038   // new shields by decoding %2E to '.'. 2048   // new shields by decoding %2E to '.'.
2039   // 2049   //
2040   // This is an optimization. Step 4 would create 2050   // This is an optimization. Step 4 would create
2041   // a new shield anyway if remove_dot_segments 2051   // a new shield anyway if remove_dot_segments
2042   // produces a problematic output, but preserving 2052   // produces a problematic output, but preserving
2043   // an existing shield avoids the memmove in 2053   // an existing shield avoids the memmove in
2044   // Step 4. 2054   // Step 4.
2045   // 2055   //
2046   // path_shield_len: number of leading bytes 2056   // path_shield_len: number of leading bytes
2047   // to skip during remove_dot_segments. 2057   // to skip during remove_dot_segments.
2048   // Always 0 (no shield) or 2 ("/." or "./"). 2058   // Always 0 (no shield) or 2 ("/." or "./").
2049   // 2059   //
HITCBC 2050   1068 auto path_shield_len = 0; 2060   1107 auto path_shield_len = 0;
HITCBC 2051   1068 if (!has_authority()) 2061   1107 if (!has_authority())
2052   { 2062   {
HITCBC 2053   316 if (p.starts_with("/./")) 2063   354 if (p.starts_with("/./"))
2054   { 2064   {
2055   // (a) Absolute path with "/." prefix. 2065   // (a) Absolute path with "/." prefix.
2056   // 2066   //
2057   // Check if "/." shields a "//" 2067   // Check if "/." shields a "//"
2058   // underneath (possibly through 2068   // underneath (possibly through
2059   // multiple "/./" layers like "/././/"). 2069   // multiple "/./" layers like "/././/").
2060   // If so, preserve the first 2 chars. 2070   // If so, preserve the first 2 chars.
2061   // 2071   //
2062   // /.//evil -> /.//evil (preserved) 2072   // /.//evil -> /.//evil (preserved)
2063   // Without it: //evil (ambiguous). 2073   // Without it: //evil (ambiguous).
2064   // /././/evil -> /.//evil (same idea) 2074   // /././/evil -> /.//evil (same idea)
2065   // 2075   //
2066   // Requires no authority (with authority, 2076   // Requires no authority (with authority,
2067   // "//" in the path is unambiguous): 2077   // "//" in the path is unambiguous):
2068   // http://h/.//x -> http://h//x (OK) 2078   // http://h/.//x -> http://h//x (OK)
2069   // Scheme is irrelevant (absolute paths 2079   // Scheme is irrelevant (absolute paths
2070   // can't be confused with a scheme): 2080   // can't be confused with a scheme):
2071   // scheme:/.//x and /.//x both need it. 2081   // scheme:/.//x and /.//x both need it.
2072   // 2082   //
HITCBC 2073   18 path_shield_len = 2; 2083   25 path_shield_len = 2;
HITCBC 2074   19 while (p.substr(path_shield_len, 3).starts_with("/./")) 2084   27 while (p.substr(path_shield_len, 3).starts_with("/./"))
2075   { 2085   {
HITCBC 2076   1 path_shield_len += 2; 2086   2 path_shield_len += 2;
2077   } 2087   }
HITCBC 2078   18 if (p.substr(path_shield_len).starts_with("//")) 2088   25 if (p.substr(path_shield_len).starts_with("//"))
2079   { 2089   {
HITCBC 2080   15 path_shield_len = 2; 2090   17 path_shield_len = 2;
2081   } 2091   }
2082   else 2092   else
2083   { 2093   {
HITCBC 2084   3 path_shield_len = 0; 2094   8 path_shield_len = 0;
2085   } 2095   }
2086   } 2096   }
HITCBC 2087   298 else if ( 2097   329 else if (
HITCBC 2088   355 !has_scheme() && 2098   417 !has_scheme() &&
HITCBC 2089   57 p.starts_with("./")) 2099   88 p.starts_with("./"))
2090   { 2100   {
2091   // (b) Relative path with "./" prefix, 2101   // (b) Relative path with "./" prefix,
2092   // no scheme. 2102   // no scheme.
2093   // 2103   //
2094   // Check if "./" shields a "//" 2104   // Check if "./" shields a "//"
2095   // underneath. If so, preserve it. 2105   // underneath. If so, preserve it.
2096   // 2106   //
2097   // .//evil -> .//evil (preserved) 2107   // .//evil -> .//evil (preserved)
2098   // Without it: //evil (ambiguous). 2108   // Without it: //evil (ambiguous).
2099   // ././/evil -> .//evil (same idea) 2109   // ././/evil -> .//evil (same idea)
2100   // 2110   //
2101   // Requires no authority (with authority, 2111   // Requires no authority (with authority,
2102   // "//" in the path is unambiguous): 2112   // "//" in the path is unambiguous):
2103   // //h/.//x -> //h//x (OK) 2113   // //h/.//x -> //h//x (OK)
2104   // Requires no scheme: with a scheme, 2114   // Requires no scheme: with a scheme,
2105   // remove_dot_segments would strip "./" 2115   // remove_dot_segments would strip "./"
2106   // and Step 4 handles any resulting "//". 2116   // and Step 4 handles any resulting "//".
2107   // 2117   //
HITCBC 2108   12 path_shield_len = 1; 2118   21 path_shield_len = 1;
HITCBC 2109   16 while (p.substr(path_shield_len, 3).starts_with("/./")) 2119   26 while (p.substr(path_shield_len, 3).starts_with("/./"))
2110   { 2120   {
HITCBC 2111   4 path_shield_len += 2; 2121   5 path_shield_len += 2;
2112   } 2122   }
HITCBC 2113   12 if (p.substr(path_shield_len).starts_with("//")) 2123   21 if (p.substr(path_shield_len).starts_with("//"))
2114   { 2124   {
HITCBC 2115   4 path_shield_len = 2; 2125   13 path_shield_len = 2;
2116   } 2126   }
2117   else 2127   else
2118   { 2128   {
HITCBC 2119   8 path_shield_len = 0; 2129   8 path_shield_len = 0;
2120   } 2130   }
2121   } 2131   }
2122   } 2132   }
2123   2133  
2124   // 2134   //
2125   // Step 3: Remove "." and ".." segments 2135   // Step 3: Remove "." and ".." segments
2126   // 2136   //
2127   // RFC 3986 Section 5.2.4. 2137   // RFC 3986 Section 5.2.4.
2128   // 2138   //
2129   // If path_shield_len > 0, the first 2139   // If path_shield_len > 0, the first
2130   // path_shield_len characters are an existing 2140   // path_shield_len characters are an existing
2131   // path shield ("/." or "./") that must 2141   // path shield ("/." or "./") that must
2132   // be preserved (see Step 2), we tell 2142   // be preserved (see Step 2), we tell
2133   // remove_dot_segments to start after that 2143   // remove_dot_segments to start after that
2134   // prefix. 2144   // prefix.
2135   // 2145   //
2136   // Note: remove_dot_segments only recognizes 2146   // Note: remove_dot_segments only recognizes
2137   // literal '.' and '..', not encoded forms like 2147   // literal '.' and '..', not encoded forms like
2138   // '%2E'. This is correct here because Step 1 2148   // '%2E'. This is correct here because Step 1
2139   // already decoded %2E to '.'. If this function 2149   // already decoded %2E to '.'. If this function
2140   // were called without Step 1, encoded dots 2150   // were called without Step 1, encoded dots
2141   // would be missed. 2151   // would be missed.
2142   // 2152   //
HITCBC 2143   1068 bool was_absolute = is_path_absolute(); 2153   1107 bool was_absolute = is_path_absolute();
HITCBC 2144   1068 p.remove_prefix(path_shield_len); 2154   1107 p.remove_prefix(path_shield_len);
HITCBC 2145   1068 p_dest += path_shield_len; 2155   1107 p_dest += path_shield_len;
HITCBC 2146   1068 auto n = detail::remove_dot_segments( 2156   1107 auto n = detail::remove_dot_segments(
HITCBC 2147   1068 p_dest, p_end, p); 2157   1107 p_dest, p_end, p);
2148   2158  
2149   // 2159   //
2150   // Step 4: Create path shield if needed, 2160   // Step 4: Create path shield if needed,
2151   // then shrink path to final size 2161   // then shrink path to final size
2152   // 2162   //
2153   // remove_dot_segments can produce output that 2163   // remove_dot_segments can produce output that
2154   // needs a 2-byte shield prefix, as explained 2164   // needs a 2-byte shield prefix, as explained
2155   // in step 2. The memmove below writes within 2165   // in step 2. The memmove below writes within
2156   // the path region; when ".." cancellation has 2166   // the path region; when ".." cancellation has
2157   // consumed >= 2 bytes there is already slack, 2167   // consumed >= 2 bytes there is already slack,
2158   // but for short inputs that take the shield 2168   // but for short inputs that take the shield
2159   // branch without any cancellation (e.g. "//"), 2169   // branch without any cancellation (e.g. "//"),
2160   // the region has to grow to fit the prefix. 2170   // the region has to grow to fit the prefix.
2161   // 2171   //
HITCBC 2162   3204 bool needs_shield = [&]() 2172   3321 bool needs_shield = [&]()
2163   { 2173   {
HITCBC 2164   1068 if (path_shield_len) 2174   1107 if (path_shield_len)
2165   { 2175   {
2166   // Step 2 already preserved a shield. 2176   // Step 2 already preserved a shield.
HITCBC 2167   19 return false; 2177   30 return false;
2168   } 2178   }
HITCBC 2169   1049 if (has_authority()) 2179   1077 if (has_authority())
2170   { 2180   {
2171   // With an authority, "//" in the path 2181   // With an authority, "//" in the path
2172   // is unambiguous and the path is always 2182   // is unambiguous and the path is always
2173   // absolute (path-abempty). 2183   // absolute (path-abempty).
HITCBC 2174   752 return false; 2184   753 return false;
2175   } 2185   }
HITCBC 2176   297 if (n == 0) 2186   324 if (n == 0)
2177   { 2187   {
2178   // Empty output. Nothing to shield. 2188   // Empty output. Nothing to shield.
HITCBC 2179   28 return false; 2189   29 return false;
2180   } 2190   }
HITCBC 2181   269 if (p_dest[0] != '/') 2191   295 if (p_dest[0] != '/')
2182   { 2192   {
2183   // Output doesn't start with '/'. 2193   // Output doesn't start with '/'.
2184   // No authority ambiguity, and if it 2194   // No authority ambiguity, and if it
2185   // was relative, it stayed relative. 2195   // was relative, it stayed relative.
HITCBC 2186   129 return false; 2196   129 return false;
2187   } 2197   }
HITCBC 2188   140 if (n >= 2 && p_dest[1] == '/') 2198   166 if (n >= 2 && p_dest[1] == '/')
2189   { 2199   {
2190   // Output starts with "//": would be 2200   // Output starts with "//": would be
2191   // misinterpreted as authority. 2201   // misinterpreted as authority.
2192   // /a/..//evil -> //evil 2202   // /a/..//evil -> //evil
2193   // so we need a shield 2203   // so we need a shield
2194   // /a/..//evil -> /.//evil 2204   // /a/..//evil -> /.//evil
HITCBC 2195   10 return true; 2205   11 return true;
2196   } 2206   }
HITCBC 2197   130 if (!was_absolute) 2207   155 if (!was_absolute)
2198   { 2208   {
2199   // Relative path became absolute: ".." 2209   // Relative path became absolute: ".."
2200   // canceled all leading segments and 2210   // canceled all leading segments and
2201   // the remaining input started with "/" 2211   // the remaining input started with "/"
2202   // because of how remove_dot_segments is 2212   // because of how remove_dot_segments is
2203   // defined. 2213   // defined.
2204   // a/..//evil -> /evil -> .//evil 2214   // a/..//evil -> /evil -> .//evil
2205   // a/b/../..//evil -> /evil -> .//evil 2215   // a/b/../..//evil -> /evil -> .//evil
HITCBC 2206   5 return true; 2216   14 return true;
2207   } 2217   }
HITCBC 2208   125 return false; 2218   141 return false;
HITCBC 2209   1068 }(); 2219   1107 }();
HITCBC 2210   1068 if (needs_shield) 2220   1107 if (needs_shield)
2211   { 2221   {
HITCBC 2212   15 if (n + 2 > pn) 2222   25 if (n + 2 > pn)
2213   { 2223   {
2214   // No ".. cancellation slack — grow the path 2224   // No ".. cancellation slack — grow the path
2215   // region to fit the 2-byte shield. p_dest may 2225   // region to fit the 2-byte shield. p_dest may
2216   // be invalidated by the underlying reallocation. 2226   // be invalidated by the underlying reallocation.
HITCBC 2217   3 p_dest = resize_impl(id_path, n + 2, op); 2227   3 p_dest = resize_impl(id_path, n + 2, op);
HITCBC 2218   3 pn = n + 2; 2228   3 pn = n + 2;
2219   } 2229   }
HITCBC 2220   15 std::memmove(p_dest + 2, p_dest, n); 2230   25 std::memmove(p_dest + 2, p_dest, n);
HITCBC 2221   15 if (was_absolute) 2231   25 if (was_absolute)
2222   { 2232   {
2223   // "/." keeps the path absolute. 2233   // "/." keeps the path absolute.
HITCBC 2224   10 p_dest[0] = '/'; 2234   11 p_dest[0] = '/';
HITCBC 2225   10 p_dest[1] = '.'; 2235   11 p_dest[1] = '.';
2226   } 2236   }
2227   else 2237   else
2228   { 2238   {
2229   // "./" keeps the path relative. 2239   // "./" keeps the path relative.
HITCBC 2230   5 p_dest[0] = '.'; 2240   14 p_dest[0] = '.';
HITCBC 2231   5 p_dest[1] = '/'; 2241   14 p_dest[1] = '/';
2232   } 2242   }
HITCBC 2233   15 path_shield_len = 2; 2243   25 path_shield_len = 2;
2234   } 2244   }
HITCBC 2235   1068 if (n != pn) 2245   1107 if (n != pn)
2236   { 2246   {
HITCBC 2237   166 BOOST_ASSERT(n < pn); 2247   200 BOOST_ASSERT(n < pn);
HITCBC 2238   166 shrink_impl(id_path, n + path_shield_len, op); 2248   200 shrink_impl(id_path, n + path_shield_len, op);
2239   } 2249   }
2240   2250  
2241   // 2251   //
2242   // Step 5: Encode colons in first segment 2252   // Step 5: Encode colons in first segment
2243   // 2253   //
2244   // If the URL has no scheme and no authority, 2254   // If the URL has no scheme and no authority,
2245   // a colon in the first path segment would be 2255   // a colon in the first path segment would be
2246   // misinterpreted as a scheme delimiter when 2256   // misinterpreted as a scheme delimiter when
2247   // serialized and re-parsed. 2257   // serialized and re-parsed.
2248   // 2258   //
2249   // Colons can appear in the first segment either 2259   // Colons can appear in the first segment either
2250   // because they were already there (decoded from 2260   // because they were already there (decoded from
2251   // %3A in Step 1), or because remove_dot_segments 2261   // %3A in Step 1), or because remove_dot_segments
2252   // (Step 3) canceled preceding segments via ".." 2262   // (Step 3) canceled preceding segments via ".."
2253   // and exposed a colon that was deeper in the path. 2263   // and exposed a colon that was deeper in the path.
2254   // 2264   //
2255   // Example (pre-existing): 2265   // Example (pre-existing):
2256   // my%3Asharona -> Step 1: my:sharona 2266   // my%3Asharona -> Step 1: my:sharona
2257   // -> Step 5: my%3Asharona 2267   // -> Step 5: my%3Asharona
2258   // Example (exposed by dot removal): 2268   // Example (exposed by dot removal):
2259   // a/../b:c -> Step 3: b:c 2269   // a/../b:c -> Step 3: b:c
2260   // -> Step 5: b%3Ac 2270   // -> Step 5: b%3Ac
2261   // 2271   //
2262   // ALL colons in the first segment must be 2272   // ALL colons in the first segment must be
2263   // encoded, not just the first one. In a 2273   // encoded, not just the first one. In a
2264   // schemeless relative-reference, the first 2274   // schemeless relative-reference, the first
2265   // segment must be segment-nz-nc (RFC 3986 2275   // segment must be segment-nz-nc (RFC 3986
2266   // Section 4.2), which does not allow ':': 2276   // Section 4.2), which does not allow ':':
2267   // path-noscheme = segment-nz-nc *( "/" segment ) 2277   // path-noscheme = segment-nz-nc *( "/" segment )
2268   // segment-nz-nc = 1*( unreserved / pct-encoded 2278   // segment-nz-nc = 1*( unreserved / pct-encoded
2269   // / sub-delims / "@" ) 2279   // / sub-delims / "@" )
2270   // So "b%3Ac:d" would fail to re-parse. 2280   // So "b%3Ac:d" would fail to re-parse.
2271   // All colons must go: "b%3Ac%3Ad". 2281   // All colons must go: "b%3Ac%3Ad".
2272   // 2282   //
2273   // Requires no scheme (with a scheme, colons 2283   // Requires no scheme (with a scheme, colons
2274   // in the path are unambiguous): 2284   // in the path are unambiguous):
2275   // scheme:a:b -> scheme:a:b (OK) 2285   // scheme:a:b -> scheme:a:b (OK)
2276   // Requires no authority (with authority, 2286   // Requires no authority (with authority,
2277   // path starts with "/" so the first segment 2287   // path starts with "/" so the first segment
2278   // is empty, no colon issue). 2288   // is empty, no colon issue).
2279   // 2289   //
HITCBC 2280   1148 if (!has_scheme() && 2290   1225 if (!has_scheme() &&
HITCBC 2281   80 !has_authority()) 2291   118 !has_authority())
2282   { 2292   {
HITCBC 2283   61 p = impl_.get(id_path); 2293   99 p = impl_.get(id_path);
HITCBC 2284   61 core::string_view first_seg = p; 2294   99 core::string_view first_seg = p;
HITCBC 2285   61 auto i = first_seg.find('/'); 2295   99 auto i = first_seg.find('/');
HITCBC 2286   61 if (i != core::string_view::npos) 2296   99 if (i != core::string_view::npos)
2287   { 2297   {
HITCBC 2288   31 first_seg = p.substr(0, i); 2298   68 first_seg = p.substr(0, i);
2289   } 2299   }
HITCBC 2290   61 if (first_seg.contains(':')) 2300   99 if (first_seg.contains(':'))
2291   { 2301   {
HITCBC 2292   13 pn = p.size(); 2302   13 pn = p.size();
2293   auto cn = 2303   auto cn =
HITCBC 2294   13 std::count( 2304   13 std::count(
2295   first_seg.begin(), 2305   first_seg.begin(),
2296   first_seg.end(), 2306   first_seg.end(),
HITCBC 2297   13 ':'); 2307   13 ':');
HITCBC 2298   13 resize_impl( 2308   13 resize_impl(
HITCBC 2299   13 id_path, pn + (2 * cn), op); 2309   13 id_path, pn + (2 * cn), op);
HITCBC 2300   13 auto begin = s_ + impl_.offset(id_path); 2310   13 auto begin = s_ + impl_.offset(id_path);
HITCBC 2301   13 auto it = begin; 2311   13 auto it = begin;
HITCBC 2302   13 auto end = begin + pn; 2312   13 auto end = begin + pn;
HITCBC 2303   103 while (it != end && 2313   103 while (it != end &&
HITCBC 2304   91 *it != '/') 2314   91 *it != '/')
2305   { 2315   {
HITCBC 2306   90 ++it; 2316   90 ++it;
2307   } 2317   }
HITCBC 2308   13 std::memmove(it + (2 * cn), it, end - it); 2318   13 std::memmove(it + (2 * cn), it, end - it);
HITCBC 2309   13 auto src = s_ + impl_.offset(id_path) + pn; 2319   13 auto src = s_ + impl_.offset(id_path) + pn;
HITCBC 2310   13 auto dest = s_ + impl_.offset(id_query); 2320   13 auto dest = s_ + impl_.offset(id_query);
HITCBC 2311   13 src -= end - it; 2321   13 src -= end - it;
HITCBC 2312   13 dest -= end - it; 2322   13 dest -= end - it;
HITCBC 2313   13 pn -= end - it; 2323   13 pn -= end - it;
2314   do { 2324   do {
HITCBC 2315   90 --src; 2325   90 --src;
HITCBC 2316   90 --dest; 2326   90 --dest;
HITCBC 2317   90 if (*src != ':') 2327   90 if (*src != ':')
2318   { 2328   {
HITCBC 2319   65 *dest = *src; 2329   65 *dest = *src;
2320   } 2330   }
2321   else 2331   else
2322   { 2332   {
HITCBC 2323   25 *dest-- = 'A'; 2333   25 *dest-- = 'A';
HITCBC 2324   25 *dest-- = '3'; 2334   25 *dest-- = '3';
HITCBC 2325   25 *dest = '%'; 2335   25 *dest = '%';
2326   } 2336   }
HITCBC 2327   90 --pn; 2337   90 --pn;
HITCBC 2328   90 } while (pn); 2338   90 } while (pn);
2329   } 2339   }
2330   } 2340   }
2331   2341  
2332   // 2342   //
2333   // Step 6: Update path parameters 2343   // Step 6: Update path parameters
2334   // 2344   //
2335   { 2345   {
  2346 + // The segment count is adjusted for the
  2347 + // "/." or "./" shield prefix, which is
  2348 + // not a segment of its own.
HITCBC 2336   1068 p = encoded_path(); 2349   1107 p = encoded_path();
HITCBC 2337 - 1068 if (p == "/") 2350 + 1107 std::size_t nseg = 0;
HITGIC 2338 - { 2351 + 1107 if (!p.empty())
DCB 2339 - 126 impl_.nseg_ = 0;  
2340 - }  
DCB 2341 - 942 else if (!p.empty())  
2342 - {  
DCB 2343 - 1628 impl_.nseg_ = detail::to_size_type(  
DCB 2344 - 814 std::count(  
DCB 2345 - 1628 p.begin() + 1, p.end(), '/') + 1);  
2346 - }  
2347 - else  
2348   { 2352   {
HITCBC 2349 - 128 impl_.nseg_ = 0; 2353 + 978 nseg = std::count(
HITGNC   2354 + 1956 p.begin() + 1, p.end(), '/') + 1;
2350   } 2355   }
HITGNC   2356 + 1107 impl_.nseg_ = detail::to_size_type(
  2357 + detail::path_segments(p, nseg));
HITCBC 2351   1068 impl_.decoded_[id_path] = 2358   1107 impl_.decoded_[id_path] =
HITCBC 2352   1068 detail::to_size_type(detail::decode_bytes_unsafe( 2359   1107 detail::to_size_type(detail::decode_bytes_unsafe(
2353   impl_.get(id_path))); 2360   impl_.get(id_path)));
2354   } 2361   }
HITCBC 2355   1068 return *this; 2362   1107 return *this;
HITCBC 2356   1068 } 2363   1107 }
2357   2364  
2358   inline 2365   inline
2359   url_base& 2366   url_base&
HITCBC 2360   221 url_base:: 2367   259 url_base::
2361   normalize_query() 2368   normalize_query()
2362   { 2369   {
HITCBC 2363   221 op_t op(*this); 2370   259 op_t op(*this);
HITCBC 2364   221 normalize_octets_impl( 2371   259 normalize_octets_impl(
2365   id_query, 2372   id_query,
2366   detail::query_chars, 2373   detail::query_chars,
2367   detail::query_ignore_chars, 2374   detail::query_ignore_chars,
2368   op); 2375   op);
HITCBC 2369   442 return *this; 2376   518 return *this;
HITCBC 2370   221 } 2377   259 }
2371   2378  
2372   inline 2379   inline
2373   url_base& 2380   url_base&
HITCBC 2374   218 url_base:: 2381   256 url_base::
2375   normalize_fragment() 2382   normalize_fragment()
2376   { 2383   {
HITCBC 2377   218 op_t op(*this); 2384   256 op_t op(*this);
HITCBC 2378   218 normalize_octets_impl( 2385   256 normalize_octets_impl(
2379   id_frag, detail::fragment_chars, op); 2386   id_frag, detail::fragment_chars, op);
HITCBC 2380   436 return *this; 2387   512 return *this;
HITCBC 2381   218 } 2388   256 }
2382   2389  
2383   inline 2390   inline
2384   url_base& 2391   url_base&
HITCBC 2385   217 url_base:: 2392   255 url_base::
2386   normalize() 2393   normalize()
2387   { 2394   {
HITCBC 2388   217 normalize_fragment(); 2395   255 normalize_fragment();
HITCBC 2389   217 normalize_query(); 2396   255 normalize_query();
HITCBC 2390   217 normalize_path(); 2397   255 normalize_path();
HITCBC 2391   217 normalize_authority(); 2398   255 normalize_authority();
HITCBC 2392   217 normalize_scheme(); 2399   255 normalize_scheme();
HITCBC 2393   217 return *this; 2400   255 return *this;
2394   } 2401   }
2395   2402  
2396   //------------------------------------------------ 2403   //------------------------------------------------
2397   // 2404   //
2398   // Implementation 2405   // Implementation
2399   // 2406   //
2400   //------------------------------------------------ 2407   //------------------------------------------------
2401   2408  
2402   inline 2409   inline
2403   void 2410   void
HITCBC 2404   43716 url_base:: 2411   44187 url_base::
2405   check_invariants() const noexcept 2412   check_invariants() const noexcept
2406   { 2413   {
HITCBC 2407   43716 BOOST_ASSERT( 2414   44187 BOOST_ASSERT(
2408   impl_.len(id_scheme) == 0 || 2415   impl_.len(id_scheme) == 0 ||
2409   impl_.get(id_scheme).ends_with(':')); 2416   impl_.get(id_scheme).ends_with(':'));
HITCBC 2410   43716 BOOST_ASSERT( 2417   44187 BOOST_ASSERT(
2411   impl_.len(id_user) == 0 || 2418   impl_.len(id_user) == 0 ||
2412   impl_.get(id_user).starts_with("//")); 2419   impl_.get(id_user).starts_with("//"));
HITCBC 2413   43716 BOOST_ASSERT( 2420   44187 BOOST_ASSERT(
2414   impl_.len(id_pass) == 0 || 2421   impl_.len(id_pass) == 0 ||
2415   impl_.get(id_user).starts_with("//")); 2422   impl_.get(id_user).starts_with("//"));
HITCBC 2416   43716 BOOST_ASSERT( 2423   44187 BOOST_ASSERT(
2417   impl_.len(id_pass) == 0 || 2424   impl_.len(id_pass) == 0 ||
2418   (impl_.len(id_pass) == 1 && 2425   (impl_.len(id_pass) == 1 &&
2419   impl_.get(id_pass) == "@") || 2426   impl_.get(id_pass) == "@") ||
2420   (impl_.len(id_pass) > 1 && 2427   (impl_.len(id_pass) > 1 &&
2421   impl_.get(id_pass).starts_with(':') && 2428   impl_.get(id_pass).starts_with(':') &&
2422   impl_.get(id_pass).ends_with('@'))); 2429   impl_.get(id_pass).ends_with('@')));
HITCBC 2423   43716 BOOST_ASSERT( 2430   44187 BOOST_ASSERT(
2424   impl_.len(id_user, id_path) == 0 || 2431   impl_.len(id_user, id_path) == 0 ||
2425   impl_.get(id_user).starts_with("//")); 2432   impl_.get(id_user).starts_with("//"));
HITCBC 2426   43716 BOOST_ASSERT(impl_.decoded_[id_path] >= 2433   44187 BOOST_ASSERT(impl_.decoded_[id_path] >=
2427   ((impl_.len(id_path) + 2) / 3)); 2434   ((impl_.len(id_path) + 2) / 3));
HITCBC 2428   43716 BOOST_ASSERT( 2435   44187 BOOST_ASSERT(
2429   impl_.len(id_port) == 0 || 2436   impl_.len(id_port) == 0 ||
2430   impl_.get(id_port).starts_with(':')); 2437   impl_.get(id_port).starts_with(':'));
HITCBC 2431   43716 BOOST_ASSERT( 2438   44187 BOOST_ASSERT(
2432   impl_.len(id_query) == 0 || 2439   impl_.len(id_query) == 0 ||
2433   impl_.get(id_query).starts_with('?')); 2440   impl_.get(id_query).starts_with('?'));
HITCBC 2434   43716 BOOST_ASSERT( 2441   44187 BOOST_ASSERT(
2435   (impl_.len(id_query) == 0 && impl_.nparam_ == 0) || 2442   (impl_.len(id_query) == 0 && impl_.nparam_ == 0) ||
2436   (impl_.len(id_query) > 0 && impl_.nparam_ > 0)); 2443   (impl_.len(id_query) > 0 && impl_.nparam_ > 0));
HITCBC 2437   43716 BOOST_ASSERT( 2444   44187 BOOST_ASSERT(
2438   impl_.len(id_frag) == 0 || 2445   impl_.len(id_frag) == 0 ||
2439   impl_.get(id_frag).starts_with('#')); 2446   impl_.get(id_frag).starts_with('#'));
HITCBC 2440   43716 BOOST_ASSERT(c_str()[size()] == '\0'); 2447   44187 BOOST_ASSERT(c_str()[size()] == '\0');
HITCBC 2441   43716 } 2448   44187 }
2442   2449  
2443   inline 2450   inline
2444   char* 2451   char*
HITCBC 2445   5534 url_base:: 2452   5544 url_base::
2446   resize_impl( 2453   resize_impl(
2447   int id, 2454   int id,
2448   std::size_t new_size, 2455   std::size_t new_size,
2449   op_t& op) 2456   op_t& op)
2450   { 2457   {
HITCBC 2451   5534 return resize_impl( 2458   5544 return resize_impl(
HITCBC 2452   5534 id, id + 1, new_size, op); 2459   5544 id, id + 1, new_size, op);
2453   } 2460   }
2454   2461  
2455   inline 2462   inline
2456   char* 2463   char*
HITCBC 2457   6260 url_base:: 2464   6279 url_base::
2458   resize_impl( 2465   resize_impl(
2459   int first, 2466   int first,
2460   int last, 2467   int last,
2461   std::size_t new_len, 2468   std::size_t new_len,
2462   op_t& op) 2469   op_t& op)
2463   { 2470   {
HITCBC 2464   6260 auto const n0 = impl_.len(first, last); 2471   6279 auto const n0 = impl_.len(first, last);
HITCBC 2465   6260 if(new_len == 0 && n0 == 0) 2472   6279 if(new_len == 0 && n0 == 0)
HITCBC 2466   1090 return s_ + impl_.offset(first); 2473   1091 return s_ + impl_.offset(first);
HITCBC 2467   5170 if(new_len <= n0) 2474   5188 if(new_len <= n0)
HITCBC 2468   2212 return shrink_impl( 2475   2220 return shrink_impl(
HITCBC 2469   2212 first, last, new_len, op); 2476   2220 first, last, new_len, op);
2470   2477  
2471   // growing 2478   // growing
HITCBC 2472   2958 std::size_t n = new_len - n0; 2479   2968 std::size_t n = new_len - n0;
HITCBC 2473   2958 reserve_impl(size() + n, op); 2480   2968 reserve_impl(size() + n, op);
2474   auto const pos = 2481   auto const pos =
HITCBC 2475   2958 impl_.offset(last); 2482   2968 impl_.offset(last);
2476   // adjust chars 2483   // adjust chars
HITCBC 2477   2958 op.move( 2484   2968 op.move(
HITCBC 2478   2958 s_ + pos + n, 2485   2968 s_ + pos + n,
HITCBC 2479   2958 s_ + pos, 2486   2968 s_ + pos,
HITCBC 2480   2958 impl_.offset(id_end) - 2487   2968 impl_.offset(id_end) -
2481   pos + 1); 2488   pos + 1);
2482   // collapse (first, last) 2489   // collapse (first, last)
HITCBC 2483   2958 impl_.collapse(first, last, 2490   2968 impl_.collapse(first, last,
HITCBC 2484   2958 impl_.offset(last) + n); 2491   2968 impl_.offset(last) + n);
2485   // shift (last, end) right 2492   // shift (last, end) right
HITCBC 2486   2958 impl_.adjust_right(last, id_end, n); 2493   2968 impl_.adjust_right(last, id_end, n);
HITCBC 2487   2958 s_[size()] = '\0'; 2494   2968 s_[size()] = '\0';
HITCBC 2488   2958 return s_ + impl_.offset(first); 2495   2968 return s_ + impl_.offset(first);
2489   } 2496   }
2490   2497  
2491   inline 2498   inline
2492   char* 2499   char*
HITCBC 2493   201 url_base:: 2500   235 url_base::
2494   shrink_impl( 2501   shrink_impl(
2495   int id, 2502   int id,
2496   std::size_t new_size, 2503   std::size_t new_size,
2497   op_t& op) 2504   op_t& op)
2498   { 2505   {
HITCBC 2499   201 return shrink_impl( 2506   235 return shrink_impl(
HITCBC 2500   201 id, id + 1, new_size, op); 2507   235 id, id + 1, new_size, op);
2501   } 2508   }
2502   2509  
2503   inline 2510   inline
2504   char* 2511   char*
HITCBC 2505   2413 url_base:: 2512   2455 url_base::
2506   shrink_impl( 2513   shrink_impl(
2507   int first, 2514   int first,
2508   int last, 2515   int last,
2509   std::size_t new_len, 2516   std::size_t new_len,
2510   op_t& op) 2517   op_t& op)
2511   { 2518   {
2512   // shrinking 2519   // shrinking
HITCBC 2513   2413 auto const n0 = impl_.len(first, last); 2520   2455 auto const n0 = impl_.len(first, last);
HITCBC 2514   2413 BOOST_ASSERT(new_len <= n0); 2521   2455 BOOST_ASSERT(new_len <= n0);
HITCBC 2515   2413 std::size_t n = n0 - new_len; 2522   2455 std::size_t n = n0 - new_len;
2516   auto const pos = 2523   auto const pos =
HITCBC 2517   2413 impl_.offset(last); 2524   2455 impl_.offset(last);
2518   // adjust chars 2525   // adjust chars
HITCBC 2519   2413 op.move( 2526   2455 op.move(
HITCBC 2520   2413 s_ + pos - n, 2527   2455 s_ + pos - n,
HITCBC 2521   2413 s_ + pos, 2528   2455 s_ + pos,
HITCBC 2522   2413 impl_.offset( 2529   2455 impl_.offset(
HITCBC 2523   2413 id_end) - pos + 1); 2530   2455 id_end) - pos + 1);
2524   // collapse (first, last) 2531   // collapse (first, last)
HITCBC 2525   2413 impl_.collapse(first, last, 2532   2455 impl_.collapse(first, last,
HITCBC 2526   2413 impl_.offset(last) - n); 2533   2455 impl_.offset(last) - n);
2527   // shift (last, end) left 2534   // shift (last, end) left
HITCBC 2528   2413 impl_.adjust_left(last, id_end, n); 2535   2455 impl_.adjust_left(last, id_end, n);
HITCBC 2529   2413 s_[size()] = '\0'; 2536   2455 s_[size()] = '\0';
HITCBC 2530   2413 return s_ + impl_.offset(first); 2537   2455 return s_ + impl_.offset(first);
2531   } 2538   }
2532   2539  
2533   //------------------------------------------------ 2540   //------------------------------------------------
2534   2541  
2535   inline 2542   inline
2536   void 2543   void
HITCBC 2537   526 url_base:: 2544   529 url_base::
2538   set_scheme_impl( 2545   set_scheme_impl(
2539   core::string_view s, 2546   core::string_view s,
2540   urls::scheme id) 2547   urls::scheme id)
2541   { 2548   {
HITCBC 2542   526 op_t op(*this, &s); 2549   529 op_t op(*this, &s);
HITCBC 2543   526 check_invariants(); 2550   529 check_invariants();
HITCBC 2544   574 grammar::parse( 2551   577 grammar::parse(
HITCBC 2545   48 s, detail::scheme_rule() 2552   48 s, detail::scheme_rule()
HITCBC 2546   574 ).value(BOOST_URL_POS); 2553   577 ).value(BOOST_URL_POS);
HITCBC 2547   478 auto const n = s.size(); 2554   481 auto const n = s.size();
HITCBC 2548   478 auto const p = impl_.offset(id_path); 2555   481 auto const p = impl_.offset(id_path);
2549   2556  
2550   // check for "./" prefix 2557   // check for "./" prefix
2551   bool const has_dot = 2558   bool const has_dot =
HITCBC 2552   1434 [this, p] 2559   1443 [this, p]
2553   { 2560   {
HITCBC 2554   478 if(impl_.nseg_ == 0) 2561   481 if(impl_.nseg_ == 0)
HITCBC 2555   213 return false; 2562   213 return false;
HITCBC 2556   265 if(first_segment().size() < 2) 2563   268 if(first_segment().size() < 2)
HITCBC 2557   42 return false; 2564   42 return false;
HITCBC 2558   223 auto const src = s_ + p; 2565   226 auto const src = s_ + p;
HITCBC 2559   223 if(src[0] != '.') 2566   226 if(src[0] != '.')
HITCBC 2560   220 return false; 2567   220 return false;
HITCBC 2561   3 if(src[1] != '/') 2568   6 if(src[1] != '/')
MISUBC 2562   ✗ return false; 2569   ✗ return false;
HITCBC 2563   3 return true; 2570   6 return true;
HITCBC 2564   478 }(); 2571   481 }();
2565   2572  
2566   // Remove "./" 2573   // Remove "./"
HITCBC 2567   478 if(has_dot) 2574   481 if(has_dot)
2568   { 2575   {
2569   // do this first, for 2576   // do this first, for
2570   // strong exception safety 2577   // strong exception safety
HITCBC 2571   6 reserve_impl( 2578   12 reserve_impl(
HITCBC 2572   3 size() + n + 1 - 2, op); 2579   6 size() + n + 1 - 2, op);
HITCBC 2573   3 op.move( 2580   6 op.move(
HITCBC 2574   3 s_ + p, 2581   6 s_ + p,
HITCBC 2575   3 s_ + p + 2, 2582   6 s_ + p + 2,
HITCBC 2576   3 size() + 1 - 2583   6 size() + 1 -
2577   (p + 2)); 2584   (p + 2));
HITCBC 2578   3 impl_.set_size( 2585   6 impl_.set_size(
2579   id_path, 2586   id_path,
HITCBC 2580   3 impl_.len(id_path) - 2); 2587   6 impl_.len(id_path) - 2);
HITGNC   2588 + 6 impl_.decoded_[id_path] -= 2;
HITCBC 2581   3 s_[size()] = '\0'; 2589   6 s_[size()] = '\0';
2582   } 2590   }
2583   2591  
HITCBC 2584   478 auto dest = resize_impl( 2592   481 auto dest = resize_impl(
2585   id_scheme, n + 1, op); 2593   id_scheme, n + 1, op);
HITCBC 2586   478 s.copy(dest, n); 2594   481 s.copy(dest, n);
HITCBC 2587   478 dest[n] = ':'; 2595   481 dest[n] = ':';
HITCBC 2588   478 impl_.scheme_ = id; 2596   481 impl_.scheme_ = id;
HITCBC 2589   478 check_invariants(); 2597   481 check_invariants();
HITCBC 2590   526 } 2598   529 }
2591   2599  
2592   inline 2600   inline
2593   char* 2601   char*
HITCBC 2594   398 url_base:: 2602   399 url_base::
2595   set_user_impl( 2603   set_user_impl(
2596   std::size_t n, 2604   std::size_t n,
2597   op_t& op) 2605   op_t& op)
2598   { 2606   {
HITCBC 2599   398 check_invariants(); 2607   399 check_invariants();
HITCBC 2600   398 if(impl_.len(id_pass) != 0) 2608   399 if(impl_.len(id_pass) != 0)
2601   { 2609   {
2602   // keep "//" 2610   // keep "//"
HITCBC 2603   147 auto dest = resize_impl( 2611   147 auto dest = resize_impl(
2604   id_user, 2 + n, op); 2612   id_user, 2 + n, op);
HITCBC 2605   147 check_invariants(); 2613   147 check_invariants();
HITCBC 2606   147 return dest + 2; 2614   147 return dest + 2;
2607   } 2615   }
2608   // add authority 2616   // add authority
2609   bool const make_absolute = 2617   bool const make_absolute =
HITCBC 2610   337 !is_path_absolute() && 2618   339 !is_path_absolute() &&
HITCBC 2611   86 !impl_.get(id_path).empty(); 2619   87 !impl_.get(id_path).empty();
HITCBC 2612   502 auto dest = resize_impl( 2620   504 auto dest = resize_impl(
HITCBC 2613   251 id_user, 2 + n + 1 + make_absolute, op); 2621   252 id_user, 2 + n + 1 + make_absolute, op);
HITCBC 2614   251 impl_.split(id_user, 2 + n); 2622   252 impl_.split(id_user, 2 + n);
HITCBC 2615   251 dest[0] = '/'; 2623   252 dest[0] = '/';
HITCBC 2616   251 dest[1] = '/'; 2624   252 dest[1] = '/';
HITCBC 2617   251 dest[2 + n] = '@'; 2625   252 dest[2 + n] = '@';
HITCBC 2618   251 if (make_absolute) 2626   252 if (make_absolute)
2619   { 2627   {
HITCBC 2620   6 impl_.split(id_pass, 1); 2628   7 impl_.split(id_pass, 1);
HITCBC 2621   6 impl_.split(id_host, 0); 2629   7 impl_.split(id_host, 0);
HITCBC 2622   6 impl_.split(id_port, 0); 2630   7 impl_.split(id_port, 0);
HITCBC 2623   6 dest[3 + n] = '/'; 2631   7 dest[3 + n] = '/';
HITGNC   2632 + 7 ++impl_.decoded_[id_path];
2624   } 2633   }
HITCBC 2625   251 check_invariants(); 2634   252 check_invariants();
HITCBC 2626   251 return dest + 2; 2635   252 return dest + 2;
2627   } 2636   }
2628   2637  
2629   inline 2638   inline
2630   char* 2639   char*
HITCBC 2631   378 url_base:: 2640   379 url_base::
2632   set_password_impl( 2641   set_password_impl(
2633   std::size_t n, 2642   std::size_t n,
2634   op_t& op) 2643   op_t& op)
2635   { 2644   {
HITCBC 2636   378 check_invariants(); 2645   379 check_invariants();
HITCBC 2637   378 if(impl_.len(id_user) != 0) 2646   379 if(impl_.len(id_user) != 0)
2638   { 2647   {
2639   // already have authority 2648   // already have authority
HITCBC 2640   314 auto const dest = resize_impl( 2649   314 auto const dest = resize_impl(
2641   id_pass, 1 + n + 1, op); 2650   id_pass, 1 + n + 1, op);
HITCBC 2642   314 dest[0] = ':'; 2651   314 dest[0] = ':';
HITCBC 2643   314 dest[n + 1] = '@'; 2652   314 dest[n + 1] = '@';
HITCBC 2644   314 check_invariants(); 2653   314 check_invariants();
HITCBC 2645   314 return dest + 1; 2654   314 return dest + 1;
2646   } 2655   }
2647   // add authority 2656   // add authority
2648   bool const make_absolute = 2657   bool const make_absolute =
HITCBC 2649   104 !is_path_absolute() && 2658   106 !is_path_absolute() &&
HITCBC 2650   40 !impl_.get(id_path).empty(); 2659   41 !impl_.get(id_path).empty();
2651   auto const dest = 2660   auto const dest =
HITCBC 2652   128 resize_impl( 2661   130 resize_impl(
2653   id_user, id_host, 2662   id_user, id_host,
HITCBC 2654   64 2 + 1 + n + 1 + make_absolute, op); 2663   65 2 + 1 + n + 1 + make_absolute, op);
HITCBC 2655   64 impl_.split(id_user, 2); 2664   65 impl_.split(id_user, 2);
HITCBC 2656   64 dest[0] = '/'; 2665   65 dest[0] = '/';
HITCBC 2657   64 dest[1] = '/'; 2666   65 dest[1] = '/';
HITCBC 2658   64 dest[2] = ':'; 2667   65 dest[2] = ':';
HITCBC 2659   64 dest[2 + n + 1] = '@'; 2668   65 dest[2 + n + 1] = '@';
HITCBC 2660   64 if (make_absolute) 2669   65 if (make_absolute)
2661   { 2670   {
HITCBC 2662   7 impl_.split(id_pass, 2 + n); 2671   8 impl_.split(id_pass, 2 + n);
HITCBC 2663   7 impl_.split(id_host, 0); 2672   8 impl_.split(id_host, 0);
HITCBC 2664   7 impl_.split(id_port, 0); 2673   8 impl_.split(id_port, 0);
HITCBC 2665   7 dest[4 + n] = '/'; 2674   8 dest[4 + n] = '/';
HITGNC   2675 + 8 ++impl_.decoded_[id_path];
2666   } 2676   }
HITCBC 2667   64 check_invariants(); 2677   65 check_invariants();
HITCBC 2668   64 return dest + 3; 2678   65 return dest + 3;
2669   } 2679   }
2670   2680  
2671   inline 2681   inline
2672   char* 2682   char*
HITCBC 2673   277 url_base:: 2683   278 url_base::
2674   set_userinfo_impl( 2684   set_userinfo_impl(
2675   std::size_t n, 2685   std::size_t n,
2676   op_t& op) 2686   op_t& op)
2677   { 2687   {
2678   // "//" {dest} "@" 2688   // "//" {dest} "@"
HITCBC 2679   277 check_invariants(); 2689   278 check_invariants();
2680   bool const make_absolute = 2690   bool const make_absolute =
HITCBC 2681   406 !is_path_absolute() && 2691   408 !is_path_absolute() &&
HITCBC 2682   129 !impl_.get(id_path).empty(); 2692   130 !impl_.get(id_path).empty();
HITCBC 2683   554 auto dest = resize_impl( 2693   556 auto dest = resize_impl(
HITCBC 2684   277 id_user, id_host, n + 3 + make_absolute, op); 2694   278 id_user, id_host, n + 3 + make_absolute, op);
HITCBC 2685   277 impl_.split(id_user, n + 2); 2695   278 impl_.split(id_user, n + 2);
HITCBC 2686   277 dest[0] = '/'; 2696   278 dest[0] = '/';
HITCBC 2687   277 dest[1] = '/'; 2697   278 dest[1] = '/';
HITCBC 2688   277 dest[n + 2] = '@'; 2698   278 dest[n + 2] = '@';
HITCBC 2689   277 if (make_absolute) 2699   278 if (make_absolute)
2690   { 2700   {
HITCBC 2691   3 impl_.split(id_pass, 1); 2701   4 impl_.split(id_pass, 1);
HITCBC 2692   3 impl_.split(id_host, 0); 2702   4 impl_.split(id_host, 0);
HITCBC 2693   3 impl_.split(id_port, 0); 2703   4 impl_.split(id_port, 0);
HITCBC 2694   3 dest[3 + n] = '/'; 2704   4 dest[3 + n] = '/';
HITGNC   2705 + 4 ++impl_.decoded_[id_path];
2695   } 2706   }
HITCBC 2696   277 check_invariants(); 2707   278 check_invariants();
HITCBC 2697   277 return dest + 2; 2708   278 return dest + 2;
2698   } 2709   }
2699   2710  
2700   inline 2711   inline
2701   char* 2712   char*
HITCBC 2702   742 url_base:: 2713   742 url_base::
2703   set_host_impl( 2714   set_host_impl(
2704   std::size_t n, 2715   std::size_t n,
2705   op_t& op) 2716   op_t& op)
2706   { 2717   {
HITCBC 2707   742 check_invariants(); 2718   742 check_invariants();
HITCBC 2708   742 if(impl_.len(id_user) == 0) 2719   742 if(impl_.len(id_user) == 0)
2709   { 2720   {
2710   // add authority 2721   // add authority
2711   bool make_absolute = 2722   bool make_absolute =
HITCBC 2712   322 !is_path_absolute() && 2723   322 !is_path_absolute() &&
HITCBC 2713   148 impl_.len(id_path) != 0; 2724   148 impl_.len(id_path) != 0;
HITCBC 2714   174 auto pn = impl_.len(id_path); 2725   174 auto pn = impl_.len(id_path);
HITCBC 2715   348 auto dest = resize_impl( 2726   348 auto dest = resize_impl(
HITCBC 2716   174 id_user, n + 2 + make_absolute, op); 2727   174 id_user, n + 2 + make_absolute, op);
HITCBC 2717   174 impl_.split(id_user, 2); 2728   174 impl_.split(id_user, 2);
HITCBC 2718   174 impl_.split(id_pass, 0); 2729   174 impl_.split(id_pass, 0);
HITCBC 2719   174 impl_.split(id_host, n); 2730   174 impl_.split(id_host, n);
HITCBC 2720   174 impl_.split(id_port, 0); 2731   174 impl_.split(id_port, 0);
HITCBC 2721   174 impl_.split(id_path, pn + make_absolute); 2732   174 impl_.split(id_path, pn + make_absolute);
HITCBC 2722   174 if (make_absolute) 2733   174 if (make_absolute)
2723   { 2734   {
HITCBC 2724   9 dest[n + 2] = '/'; 2735   9 dest[n + 2] = '/';
HITCBC 2725   9 ++impl_.decoded_[id_path]; 2736   9 ++impl_.decoded_[id_path];
2726   } 2737   }
HITCBC 2727   174 dest[0] = '/'; 2738   174 dest[0] = '/';
HITCBC 2728   174 dest[1] = '/'; 2739   174 dest[1] = '/';
HITCBC 2729   174 check_invariants(); 2740   174 check_invariants();
HITCBC 2730   174 return dest + 2; 2741   174 return dest + 2;
2731   } 2742   }
2732   // already have authority 2743   // already have authority
HITCBC 2733   568 auto const dest = resize_impl( 2744   568 auto const dest = resize_impl(
2734   id_host, n, op); 2745   id_host, n, op);
HITCBC 2735   568 check_invariants(); 2746   568 check_invariants();
HITCBC 2736   568 return dest; 2747   568 return dest;
2737   } 2748   }
2738   2749  
2739   inline 2750   inline
2740   char* 2751   char*
HITCBC 2741   605 url_base:: 2752   605 url_base::
2742   set_port_impl( 2753   set_port_impl(
2743   std::size_t n, 2754   std::size_t n,
2744   op_t& op) 2755   op_t& op)
2745   { 2756   {
HITCBC 2746   605 check_invariants(); 2757   605 check_invariants();
HITCBC 2747   605 if(impl_.len(id_user) != 0) 2758   605 if(impl_.len(id_user) != 0)
2748   { 2759   {
2749   // authority exists 2760   // authority exists
HITCBC 2750   504 auto dest = resize_impl( 2761   504 auto dest = resize_impl(
2751   id_port, n + 1, op); 2762   id_port, n + 1, op);
HITCBC 2752   504 dest[0] = ':'; 2763   504 dest[0] = ':';
HITCBC 2753   504 check_invariants(); 2764   504 check_invariants();
HITCBC 2754   504 return dest + 1; 2765   504 return dest + 1;
2755   } 2766   }
2756   bool make_absolute = 2767   bool make_absolute =
HITCBC 2757   174 !is_path_absolute() && 2768   174 !is_path_absolute() &&
HITCBC 2758   73 impl_.len(id_path) != 0; 2769   73 impl_.len(id_path) != 0;
HITCBC 2759   202 auto dest = resize_impl( 2770   202 auto dest = resize_impl(
HITCBC 2760   101 id_user, 3 + n + make_absolute, op); 2771   101 id_user, 3 + n + make_absolute, op);
HITCBC 2761   101 impl_.split(id_user, 2); 2772   101 impl_.split(id_user, 2);
HITCBC 2762   101 impl_.split(id_pass, 0); 2773   101 impl_.split(id_pass, 0);
HITCBC 2763   101 impl_.split(id_host, 0); 2774   101 impl_.split(id_host, 0);
HITCBC 2764   101 dest[0] = '/'; 2775   101 dest[0] = '/';
HITCBC 2765   101 dest[1] = '/'; 2776   101 dest[1] = '/';
HITCBC 2766   101 dest[2] = ':'; 2777   101 dest[2] = ':';
HITCBC 2767   101 if (make_absolute) 2778   101 if (make_absolute)
2768   { 2779   {
HITCBC 2769   8 impl_.split(id_port, n + 1); 2780   8 impl_.split(id_port, n + 1);
HITCBC 2770   8 dest[n + 3] = '/'; 2781   8 dest[n + 3] = '/';
HITCBC 2771   8 ++impl_.decoded_[id_path]; 2782   8 ++impl_.decoded_[id_path];
2772   } 2783   }
HITCBC 2773   101 check_invariants(); 2784   101 check_invariants();
HITCBC 2774   101 return dest + 3; 2785   101 return dest + 3;
2775   } 2786   }
2776   2787  
2777   inline 2788   inline
2778   char* 2789   char*
HITCBC 2779   597 url_base:: 2790   598 url_base::
2780   set_path_impl( 2791   set_path_impl(
2781   std::size_t n, 2792   std::size_t n,
2782   op_t& op) 2793   op_t& op)
2783   { 2794   {
HITCBC 2784   597 check_invariants(); 2795   598 check_invariants();
HITCBC 2785   597 auto const dest = resize_impl( 2796   598 auto const dest = resize_impl(
2786   id_path, n, op); 2797   id_path, n, op);
HITCBC 2787   597 return dest; 2798   598 return dest;
2788   } 2799   }
2789   2800  
2790   2801  
2791   //------------------------------------------------ 2802   //------------------------------------------------
2792   2803  
2793   // return the first segment of the path. 2804   // return the first segment of the path.
2794   // this is needed for some algorithms. 2805   // this is needed for some algorithms.
2795   inline 2806   inline
2796   core::string_view 2807   core::string_view
HITCBC 2797   303 url_base:: 2808   268 url_base::
2798   first_segment() const noexcept 2809   first_segment() const noexcept
2799   { 2810   {
HITCBC 2800   303 if(impl_.nseg_ == 0) 2811   268 if(impl_.nseg_ == 0)
MISLBC 2801   9 return {}; 2812   ✗ return {};
HITCBC 2802   294 auto const p0 = impl_.cs_ + 2813   268 auto const p0 = impl_.cs_ +
HITCBC 2803   294 impl_.offset(id_path) + 2814   268 impl_.offset(id_path) +
HITCBC 2804   294 detail::path_prefix( 2815   268 detail::path_prefix(
HITCBC 2805   294 impl_.get(id_path)); 2816   268 impl_.get(id_path));
HITCBC 2806   294 auto const end = impl_.cs_ + 2817   268 auto const end = impl_.cs_ +
HITCBC 2807   294 impl_.offset(id_query); 2818   268 impl_.offset(id_query);
HITCBC 2808   294 if(impl_.nseg_ == 1) 2819   268 if(impl_.nseg_ == 1)
HITCBC 2809   182 return core::string_view( 2820   160 return core::string_view(
HITCBC 2810   91 p0, end - p0); 2821   80 p0, end - p0);
HITCBC 2811   203 auto p = p0; 2822   188 auto p = p0;
HITCBC 2812   910 while(*p != '/') 2823   855 while(*p != '/')
2813   { 2824   {
HITCBC 2814   707 BOOST_ASSERT(p < end); 2825   667 BOOST_ASSERT(p < end);
HITCBC 2815   707 ++p; 2826   667 ++p;
2816   } 2827   }
HITCBC 2817   203 return core::string_view(p0, p - p0); 2828   188 return core::string_view(p0, p - p0);
2818   } 2829   }
2819   2830  
2820   inline 2831   inline
2821   detail::segments_iter_impl 2832   detail::segments_iter_impl
HITCBC 2822   2315 url_base:: 2833   2315 url_base::
2823   edit_segments( 2834   edit_segments(
2824   detail::segments_iter_impl const& it0, 2835   detail::segments_iter_impl const& it0,
2825   detail::segments_iter_impl const& it1, 2836   detail::segments_iter_impl const& it1,
2826   detail::any_segments_iter&& src, 2837   detail::any_segments_iter&& src,
2827   // -1 = preserve 2838   // -1 = preserve
2828   // 0 = make relative (can fail) 2839   // 0 = make relative (can fail)
2829   // 1 = make absolute 2840   // 1 = make absolute
2830   int absolute) 2841   int absolute)
2831   { 2842   {
2832   // Iterator doesn't belong to this url 2843   // Iterator doesn't belong to this url
HITCBC 2833   2315 BOOST_ASSERT(it0.ref.alias_of(impl_)); 2844   2315 BOOST_ASSERT(it0.ref.alias_of(impl_));
2834   2845  
2835   // Iterator doesn't belong to this url 2846   // Iterator doesn't belong to this url
HITCBC 2836   2315 BOOST_ASSERT(it1.ref.alias_of(impl_)); 2847   2315 BOOST_ASSERT(it1.ref.alias_of(impl_));
2837   2848  
2838   // Iterator is in the wrong order 2849   // Iterator is in the wrong order
HITCBC 2839   2315 BOOST_ASSERT(it0.index <= it1.index); 2850   2315 BOOST_ASSERT(it0.index <= it1.index);
2840   2851  
2841   // Iterator is out of range 2852   // Iterator is out of range
HITCBC 2842   2315 BOOST_ASSERT(it0.index <= impl_.nseg_); 2853   2315 BOOST_ASSERT(it0.index <= impl_.nseg_);
HITCBC 2843   2315 BOOST_ASSERT(it0.pos <= impl_.len(id_path)); 2854   2315 BOOST_ASSERT(it0.pos <= impl_.len(id_path));
2844   2855  
2845   // Iterator is out of range 2856   // Iterator is out of range
HITCBC 2846   2315 BOOST_ASSERT(it1.index <= impl_.nseg_); 2857   2315 BOOST_ASSERT(it1.index <= impl_.nseg_);
HITCBC 2847   2315 BOOST_ASSERT(it1.pos <= impl_.len(id_path)); 2858   2315 BOOST_ASSERT(it1.pos <= impl_.len(id_path));
2848   2859  
2849   //------------------------------------------------ 2860   //------------------------------------------------
2850   // 2861   //
2851   // Calculate output prefix 2862   // Calculate output prefix
2852   // 2863   //
2853   // 0 = "" 2864   // 0 = ""
2854   // 1 = "/" 2865   // 1 = "/"
2855   // 2 = "./" 2866   // 2 = "./"
2856   // 3 = "/./" 2867   // 3 = "/./"
2857   // 2868   //
HITCBC 2858   2315 bool const is_abs = is_path_absolute(); 2869   2315 bool const is_abs = is_path_absolute();
HITCBC 2859   2315 if(has_authority()) 2870   2315 if(has_authority())
2860   { 2871   {
2861   // Check if the new 2872   // Check if the new
2862   // path would be empty 2873   // path would be empty
HITCBC 2863   1855 if( src.fast_nseg == 0 && 2874   1855 if( src.fast_nseg == 0 &&
HITCBC 2864   911 it0.index == 0 && 2875   911 it0.index == 0 &&
HITCBC 2865   655 it1.index == impl_.nseg_) 2876   655 it1.index == impl_.nseg_)
2866   { 2877   {
2867   // VFALCO we don't have 2878   // VFALCO we don't have
2868   // access to nchar this early 2879   // access to nchar this early
2869   // 2880   //
2870   //BOOST_ASSERT(nchar == 0); 2881   //BOOST_ASSERT(nchar == 0);
HITCBC 2871   627 absolute = 0; 2882   627 absolute = 0;
2872   } 2883   }
2873   else 2884   else
2874   { 2885   {
2875   // prefix "/" required 2886   // prefix "/" required
HITCBC 2876   1228 absolute = 1; 2887   1228 absolute = 1;
2877   } 2888   }
2878   } 2889   }
HITCBC 2879   460 else if(absolute < 0) 2890   460 else if(absolute < 0)
2880   { 2891   {
HITCBC 2881   460 absolute = is_abs; // preserve 2892   460 absolute = is_abs; // preserve
2882   } 2893   }
HITCBC 2883   2315 auto const path_pos = impl_.offset(id_path); 2894   2315 auto const path_pos = impl_.offset(id_path);
2884   2895  
HITCBC 2885   2315 std::size_t nchar = 0; 2896   2315 std::size_t nchar = 0;
HITCBC 2886   2315 std::size_t prefix = 0; 2897   2315 std::size_t prefix = 0;
HITCBC 2887   2315 bool encode_colons = false; 2898   2315 bool encode_colons = false;
HITCBC 2888   2315 bool cp_src_prefix = false; 2899   2315 bool cp_src_prefix = false;
HITCBC 2889   2315 if(it0.index > 0) 2900   2315 if(it0.index > 0)
2890   { 2901   {
2891   // first segment unchanged 2902   // first segment unchanged
HITCBC 2892   870 prefix = src.fast_nseg > 0; 2903   870 prefix = src.fast_nseg > 0;
2893   } 2904   }
HITCBC 2894   1445 else if(src.fast_nseg > 0) 2905   1445 else if(src.fast_nseg > 0)
2895   { 2906   {
2896   // first segment from src 2907   // first segment from src
HITCBC 2897   718 if(! src.front.empty()) 2908   718 if(! src.front.empty())
2898   { 2909   {
HITCBC 2899   640 if( src.front == "." && 2910   640 if( src.front == "." &&
HITCBC 2900   7 src.fast_nseg > 1) 2911   7 src.fast_nseg > 1)
HITCBC 2901   4 if (src.s.empty()) 2912   4 if (src.s.empty())
2902   { 2913   {
2903   // if front is ".", we need the extra "." in the prefix 2914   // if front is ".", we need the extra "." in the prefix
2904   // which will maintain the invariant that segments represent 2915   // which will maintain the invariant that segments represent
2905   // {"."} 2916   // {"."}
HITCBC 2906   4 prefix = 2 + absolute; 2917   4 prefix = 2 + absolute;
2907   } 2918   }
2908   else 2919   else
2909   { 2920   {
2910   // if the "." prefix is explicitly required from set_path 2921   // if the "." prefix is explicitly required from set_path
2911   // we do not include an extra "." segment 2922   // we do not include an extra "." segment
MISUBC 2912   ✗ prefix = absolute; 2923   ✗ prefix = absolute;
MISUBC 2913   ✗ cp_src_prefix = true; 2924   ✗ cp_src_prefix = true;
2914   } 2925   }
HITCBC 2915   629 else if(absolute) 2926   629 else if(absolute)
HITCBC 2916   542 prefix = 1; 2927   542 prefix = 1;
HITCBC 2917   166 else if(has_scheme() || 2928   166 else if(has_scheme() ||
HITCBC 2918   79 ! src.front.contains(':')) 2929   79 ! src.front.contains(':'))
HITCBC 2919   82 prefix = 0; 2930   82 prefix = 0;
2920   else 2931   else
2921   { 2932   {
HITCBC 2922   5 prefix = 0; 2933   5 prefix = 0;
HITCBC 2923   5 encode_colons = true; 2934   5 encode_colons = true;
2924   } 2935   }
2925   } 2936   }
2926   else 2937   else
2927   { 2938   {
HITCBC 2928   85 prefix = 2 + absolute; 2939   85 prefix = 2 + absolute;
2929   } 2940   }
2930   } 2941   }
2931   else 2942   else
2932   { 2943   {
2933   // first segment from it1 2944   // first segment from it1
HITCBC 2934   727 auto const p = 2945   727 auto const p =
HITCBC 2935   727 impl_.cs_ + path_pos + it1.pos; 2946   727 impl_.cs_ + path_pos + it1.pos;
HITCBC 2936   1454 switch(impl_.cs_ + 2947   1454 switch(impl_.cs_ +
HITCBC 2937   727 impl_.offset(id_query) - p) 2948   727 impl_.offset(id_query) - p)
2938   { 2949   {
HITCBC 2939   680 case 0: 2950   680 case 0:
2940   // points to end 2951   // points to end
HITCBC 2941   680 prefix = absolute; 2952   680 prefix = absolute;
HITCBC 2942   680 break; 2953   680 break;
HITCBC 2943   36 default: 2954   36 default:
HITCBC 2944   36 BOOST_ASSERT(*p == '/'); 2955   36 BOOST_ASSERT(*p == '/');
HITCBC 2945   36 if(p[1] != '/') 2956   36 if(p[1] != '/')
2946   { 2957   {
HITCBC 2947   35 if(absolute) 2958   35 if(absolute)
HITCBC 2948   29 prefix = 1; 2959   29 prefix = 1;
HITCBC 2949   11 else if(has_scheme() || 2960   11 else if(has_scheme() ||
HITCBC 2950   11 ! it1.dereference().contains(':')) 2961   11 ! it1.dereference().contains(':'))
HITCBC 2951   5 prefix = 0; 2962   5 prefix = 0;
2952   else 2963   else
HITCBC 2953   1 prefix = 2; 2964   1 prefix = 2;
HITCBC 2954   35 break; 2965   35 break;
2955   } 2966   }
2956   // empty 2967   // empty
2957   BOOST_FALLTHROUGH; 2968   BOOST_FALLTHROUGH;
2958   case 1: 2969   case 1:
2959   // empty 2970   // empty
HITCBC 2960   12 BOOST_ASSERT(*p == '/'); 2971   12 BOOST_ASSERT(*p == '/');
HITCBC 2961   12 prefix = 2 + absolute; 2972   12 prefix = 2 + absolute;
HITCBC 2962   12 break; 2973   12 break;
2963   } 2974   }
2964   } 2975   }
2965   2976  
2966   // append '/' to new segs 2977   // append '/' to new segs
2967   // if inserting at front. 2978   // if inserting at front.
HITCBC 2968   2315 std::size_t const suffix = 2979   2315 std::size_t const suffix =
HITCBC 2969   3234 it1.index == 0 && 2980   3234 it1.index == 0 &&
HITCBC 2970   2494 impl_.nseg_ > 0 && 2981   2494 impl_.nseg_ > 0 &&
HITCBC 2971   179 src.fast_nseg > 0; 2982   179 src.fast_nseg > 0;
2972   2983  
2973   //------------------------------------------------ 2984   //------------------------------------------------
2974   // 2985   //
2975   // Measure the number of encoded characters 2986   // Measure the number of encoded characters
2976   // of output, and the number of inserted 2987   // of output, and the number of inserted
2977   // segments including internal separators. 2988   // segments including internal separators.
2978   // 2989   //
HITCBC 2979   2315 src.encode_colons = encode_colons; 2990   2315 src.encode_colons = encode_colons;
HITCBC 2980   2315 std::size_t nseg = 0; 2991   2315 std::size_t nseg = 0;
HITCBC 2981   2315 if(src.measure(nchar)) 2992   2315 if(src.measure(nchar))
2982   { 2993   {
HITCBC 2983   1279 src.encode_colons = false; 2994   1279 src.encode_colons = false;
2984   for(;;) 2995   for(;;)
2985   { 2996   {
HITCBC 2986   1605 ++nseg; 2997   1605 ++nseg;
HITCBC 2987   1605 if(! src.measure(nchar)) 2998   1605 if(! src.measure(nchar))
HITCBC 2988   1277 break; 2999   1277 break;
HITCBC 2989   326 ++nchar; 3000   326 ++nchar;
2990   } 3001   }
2991   } 3002   }
2992   3003  
HITCBC 2993   2313 switch(src.fast_nseg) 3004   2313 switch(src.fast_nseg)
2994   { 3005   {
HITCBC 2995   1036 case 0: 3006   1036 case 0:
HITCBC 2996   1036 BOOST_ASSERT(nseg == 0); 3007   1036 BOOST_ASSERT(nseg == 0);
HITCBC 2997   1036 break; 3008   1036 break;
HITCBC 2998   1089 case 1: 3009   1089 case 1:
HITCBC 2999   1089 BOOST_ASSERT(nseg == 1); 3010   1089 BOOST_ASSERT(nseg == 1);
HITCBC 3000   1089 break; 3011   1089 break;
HITCBC 3001   188 case 2: 3012   188 case 2:
HITCBC 3002   188 BOOST_ASSERT(nseg >= 2); 3013   188 BOOST_ASSERT(nseg >= 2);
HITCBC 3003   188 break; 3014   188 break;
3004   } 3015   }
3005   3016  
3006   //------------------------------------------------ 3017   //------------------------------------------------
3007   // 3018   //
3008   // Calculate [pos0, pos1) to remove 3019   // Calculate [pos0, pos1) to remove
3009   // 3020   //
HITCBC 3010   2313 auto pos0 = it0.pos; 3021   2313 auto pos0 = it0.pos;
HITCBC 3011   2313 if(it0.index == 0) 3022   2313 if(it0.index == 0)
3012   { 3023   {
3013   // patch pos for prefix 3024   // patch pos for prefix
HITCBC 3014   1443 pos0 = 0; 3025   1443 pos0 = 0;
3015   } 3026   }
HITCBC 3016   2313 auto pos1 = it1.pos; 3027   2313 auto pos1 = it1.pos;
HITCBC 3017   2313 if(it1.index == 0) 3028   2313 if(it1.index == 0)
3018   { 3029   {
3019   // patch pos for prefix 3030   // patch pos for prefix
HITCBC 3020   919 pos1 = detail::path_prefix( 3031   919 pos1 = detail::path_prefix(
3021   impl_.get(id_path)); 3032   impl_.get(id_path));
3022   } 3033   }
HITCBC 3023   1394 else if( 3034   1394 else if(
HITCBC 3024   1394 it0.index == 0 && 3035   1394 it0.index == 0 &&
HITCBC 3025   524 it1.index < impl_.nseg_ && 3036   524 it1.index < impl_.nseg_ &&
3026   nseg == 0) 3037   nseg == 0)
3027   { 3038   {
3028   // Remove the slash from segment it1 3039   // Remove the slash from segment it1
3029   // if it is becoming the new first 3040   // if it is becoming the new first
3030   // segment. 3041   // segment.
HITCBC 3031   47 ++pos1; 3042   47 ++pos1;
3032   } 3043   }
3033   // calc decoded size of old range 3044   // calc decoded size of old range
3034   auto const dn0 = 3045   auto const dn0 =
HITCBC 3035   2313 detail::decode_bytes_unsafe( 3046   2313 detail::decode_bytes_unsafe(
3036   core::string_view( 3047   core::string_view(
HITCBC 3037   2313 impl_.cs_ + 3048   2313 impl_.cs_ +
HITCBC 3038   2313 impl_.offset(id_path) + 3049   2313 impl_.offset(id_path) +
3039   pos0, 3050   pos0,
3040   pos1 - pos0)); 3051   pos1 - pos0));
3041   3052  
3042   //------------------------------------------------ 3053   //------------------------------------------------
3043   // 3054   //
3044   // Resize 3055   // Resize
3045   // 3056   //
HITCBC 3046   4626 op_t op(*this, &src.s); 3057   4626 op_t op(*this, &src.s);
3047   char* dest; 3058   char* dest;
3048   char const* end; 3059   char const* end;
3049   { 3060   {
HITCBC 3050   2313 auto const nremove = pos1 - pos0; 3061   2313 auto const nremove = pos1 - pos0;
3051   // check overflow 3062   // check overflow
HITCBC 3052   4626 if( nchar <= max_size() && ( 3063   4626 if( nchar <= max_size() && (
HITCBC 3053   2313 prefix + suffix <= 3064   2313 prefix + suffix <=
HITCBC 3054   2313 max_size() - nchar)) 3065   2313 max_size() - nchar))
3055   { 3066   {
HITCBC 3056   2313 nchar = prefix + nchar + suffix; 3067   2313 nchar = prefix + nchar + suffix;
HITCBC 3057   3484 if( nchar <= nremove || 3068   3484 if( nchar <= nremove ||
HITCBC 3058   1171 nchar - nremove <= 3069   1171 nchar - nremove <=
HITCBC 3059   1171 max_size() - size()) 3070   1171 max_size() - size())
HITCBC 3060   2313 goto ok; 3071   2313 goto ok;
3061   } 3072   }
3062   // too large 3073   // too large
MISUBC 3063   ✗ detail::throw_length_error(); 3074   ✗ detail::throw_length_error();
HITCBC 3064   2313 ok: 3075   2313 ok:
3065   auto const new_size = 3076   auto const new_size =
HITCBC 3066   2313 size() + nchar - nremove; 3077   2313 size() + nchar - nremove;
HITCBC 3067   2313 reserve_impl(new_size, op); 3078   2313 reserve_impl(new_size, op);
HITCBC 3068   2313 dest = s_ + path_pos + pos0; 3079   2313 dest = s_ + path_pos + pos0;
HITCBC 3069   2313 op.move( 3080   2313 op.move(
HITCBC 3070   2313 dest + nchar, 3081   2313 dest + nchar,
HITCBC 3071   2313 s_ + path_pos + pos1, 3082   2313 s_ + path_pos + pos1,
HITCBC 3072   2313 size() - path_pos - pos1); 3083   2313 size() - path_pos - pos1);
HITCBC 3073   4626 impl_.set_size( 3084   4626 impl_.set_size(
3074   id_path, 3085   id_path,
HITCBC 3075   2313 impl_.len(id_path) + nchar - nremove); 3086   2313 impl_.len(id_path) + nchar - nremove);
HITCBC 3076   2313 BOOST_ASSERT(size() == new_size); 3087   2313 BOOST_ASSERT(size() == new_size);
HITCBC 3077   2313 end = dest + nchar; 3088   2313 end = dest + nchar;
HITCBC 3078   2313 auto const nseg1 = 3089   2313 auto const nseg1 =
HITCBC 3079   2313 static_cast<std::ptrdiff_t>(impl_.nseg_) + 3090   2313 static_cast<std::ptrdiff_t>(impl_.nseg_) +
HITCBC 3080   2313 static_cast<std::ptrdiff_t>(nseg) - 3091   2313 static_cast<std::ptrdiff_t>(nseg) -
HITCBC 3081   2313 static_cast<std::ptrdiff_t>(it1.index) + 3092   2313 static_cast<std::ptrdiff_t>(it1.index) +
HITCBC 3082   2313 static_cast<std::ptrdiff_t>(it0.index) - 3093   2313 static_cast<std::ptrdiff_t>(it0.index) -
3083   static_cast<std::ptrdiff_t>(cp_src_prefix); 3094   static_cast<std::ptrdiff_t>(cp_src_prefix);
HITCBC 3084   2313 BOOST_ASSERT(nseg1 >= 0); 3095   2313 BOOST_ASSERT(nseg1 >= 0);
HITCBC 3085   2313 impl_.nseg_ = detail::to_size_type(nseg1); 3096   2313 impl_.nseg_ = detail::to_size_type(nseg1);
HITCBC 3086   2313 if(s_) 3097   2313 if(s_)
HITCBC 3087   2306 s_[size()] = '\0'; 3098   2306 s_[size()] = '\0';
3088   } 3099   }
3089   3100  
3090   //------------------------------------------------ 3101   //------------------------------------------------
3091   // 3102   //
3092   // Output segments and internal separators: 3103   // Output segments and internal separators:
3093   // 3104   //
3094   // prefix [ segment [ '/' segment ] ] suffix 3105   // prefix [ segment [ '/' segment ] ] suffix
3095   // 3106   //
HITCBC 3096   2313 auto const dest0 = dest; 3107   2313 auto const dest0 = dest;
HITCBC 3097   2313 switch(prefix) 3108   2313 switch(prefix)
3098   { 3109   {
HITCBC 3099   60 case 3: 3110   60 case 3:
HITCBC 3100   60 *dest++ = '/'; 3111   60 *dest++ = '/';
HITCBC 3101   60 *dest++ = '.'; 3112   60 *dest++ = '.';
HITCBC 3102   60 *dest++ = '/'; 3113   60 *dest++ = '/';
HITCBC 3103   60 break; 3114   60 break;
HITCBC 3104   42 case 2: 3115   42 case 2:
HITCBC 3105   42 *dest++ = '.'; 3116   42 *dest++ = '.';
3106   BOOST_FALLTHROUGH; 3117   BOOST_FALLTHROUGH;
HITCBC 3107   1193 case 1: 3118   1193 case 1:
HITCBC 3108   1193 *dest++ = '/'; 3119   1193 *dest++ = '/';
HITCBC 3109   1193 break; 3120   1193 break;
HITCBC 3110   1060 default: 3121   1060 default:
HITCBC 3111   1060 break; 3122   1060 break;
3112   } 3123   }
HITCBC 3113   2313 src.rewind(); 3124   2313 src.rewind();
HITCBC 3114   2313 if(nseg > 0) 3125   2313 if(nseg > 0)
3115   { 3126   {
HITCBC 3116   1277 src.encode_colons = encode_colons; 3127   1277 src.encode_colons = encode_colons;
3117   for(;;) 3128   for(;;)
3118   { 3129   {
HITCBC 3119   1603 src.copy(dest, end); 3130   1603 src.copy(dest, end);
HITCBC 3120   1603 if(--nseg == 0) 3131   1603 if(--nseg == 0)
HITCBC 3121   1277 break; 3132   1277 break;
HITCBC 3122   326 *dest++ = '/'; 3133   326 *dest++ = '/';
HITCBC 3123   326 src.encode_colons = false; 3134   326 src.encode_colons = false;
3124   } 3135   }
HITCBC 3125   1277 if(suffix) 3136   1277 if(suffix)
HITCBC 3126   179 *dest++ = '/'; 3137   179 *dest++ = '/';
3127   } 3138   }
HITCBC 3128   2313 BOOST_ASSERT(dest == dest0 + nchar); 3139   2313 BOOST_ASSERT(dest == dest0 + nchar);
3129   3140  
3130   // calc decoded size of new range, 3141   // calc decoded size of new range,
3131   auto const dn = 3142   auto const dn =
HITCBC 3132   2313 detail::decode_bytes_unsafe( 3143   2313 detail::decode_bytes_unsafe(
HITCBC 3133   2313 core::string_view(dest0, dest - dest0)); 3144   2313 core::string_view(dest0, dest - dest0));
HITCBC 3134   2313 if(dn >= dn0) 3145   2313 if(dn >= dn0)
HITCBC 3135   1401 impl_.decoded_[id_path] += 3146   1401 impl_.decoded_[id_path] +=
HITCBC 3136   1401 detail::to_size_type(dn - dn0); 3147   1401 detail::to_size_type(dn - dn0);
3137   else 3148   else
HITCBC 3138   912 impl_.decoded_[id_path] -= 3149   912 impl_.decoded_[id_path] -=
HITCBC 3139   912 detail::to_size_type(dn0 - dn); 3150   912 detail::to_size_type(dn0 - dn);
3140   3151  
3141   return detail::segments_iter_impl( 3152   return detail::segments_iter_impl(
HITCBC 3142   4626 impl_, pos0, it0.index); 3153   4626 impl_, pos0, it0.index);
3143   } 3154   }
3144   3155  
3145   //------------------------------------------------ 3156   //------------------------------------------------
3146   3157  
3147   inline 3158   inline
3148   auto 3159   auto
HITCBC 3149   1701 url_base:: 3160   1701 url_base::
3150   edit_params( 3161   edit_params(
3151   detail::params_iter_impl const& it0, 3162   detail::params_iter_impl const& it0,
3152   detail::params_iter_impl const& it1, 3163   detail::params_iter_impl const& it1,
3153   detail::any_params_iter&& src) -> 3164   detail::any_params_iter&& src) ->
3154   detail::params_iter_impl 3165   detail::params_iter_impl
3155   { 3166   {
HITCBC 3156   1701 auto pos0 = impl_.offset(id_query); 3167   1701 auto pos0 = impl_.offset(id_query);
HITCBC 3157   1701 auto pos1 = pos0 + it1.pos; 3168   1701 auto pos1 = pos0 + it1.pos;
HITCBC 3158   1701 pos0 = pos0 + it0.pos; 3169   1701 pos0 = pos0 + it0.pos;
3159   3170  
3160   // Iterators belong to this url 3171   // Iterators belong to this url
HITCBC 3161   1701 BOOST_ASSERT(it0.ref.alias_of(impl_)); 3172   1701 BOOST_ASSERT(it0.ref.alias_of(impl_));
HITCBC 3162   1701 BOOST_ASSERT(it1.ref.alias_of(impl_)); 3173   1701 BOOST_ASSERT(it1.ref.alias_of(impl_));
3163   3174  
3164   // Iterators is in the right order 3175   // Iterators is in the right order
HITCBC 3165   1701 BOOST_ASSERT(it0.index <= it1.index); 3176   1701 BOOST_ASSERT(it0.index <= it1.index);
3166   3177  
3167   // Iterators are within range 3178   // Iterators are within range
HITCBC 3168   1701 BOOST_ASSERT(it0.index <= impl_.nparam_); 3179   1701 BOOST_ASSERT(it0.index <= impl_.nparam_);
HITCBC 3169   1701 BOOST_ASSERT(pos0 <= impl_.offset(id_frag)); 3180   1701 BOOST_ASSERT(pos0 <= impl_.offset(id_frag));
HITCBC 3170   1701 BOOST_ASSERT(it1.index <= impl_.nparam_); 3181   1701 BOOST_ASSERT(it1.index <= impl_.nparam_);
HITCBC 3171   1701 BOOST_ASSERT(pos1 <= impl_.offset(id_frag)); 3182   1701 BOOST_ASSERT(pos1 <= impl_.offset(id_frag));
3172   3183  
3173   // calc decoded size of old range, 3184   // calc decoded size of old range,
3174   // minus one for the leading '?' which is 3185   // minus one for the leading '?' which is
3175   // not counted in decoded_[id_query]. 3186   // not counted in decoded_[id_query].
3176   // dn0 may be -1 here when the old range is 3187   // dn0 may be -1 here when the old range is
3177   // empty and the query was non-empty; the 3188   // empty and the query was non-empty; the
3178   // matching subtraction on dn below cancels 3189   // matching subtraction on dn below cancels
3179   // that out when the delta is taken. 3190   // that out when the delta is taken.
3180   auto dn0 = 3191   auto dn0 =
3181   static_cast<std::ptrdiff_t>( 3192   static_cast<std::ptrdiff_t>(
HITCBC 3182   1701 detail::decode_bytes_unsafe( 3193   1701 detail::decode_bytes_unsafe(
3183   core::string_view( 3194   core::string_view(
HITCBC 3184   1701 impl_.cs_ + pos0, 3195   1701 impl_.cs_ + pos0,
HITCBC 3185   1701 pos1 - pos0))); 3196   1701 pos1 - pos0)));
HITCBC 3186   1701 if(impl_.len(id_query) > 0) 3197   1701 if(impl_.len(id_query) > 0)
HITCBC 3187   1162 dn0 -= 1; 3198   1162 dn0 -= 1;
3188   3199  
3189   //------------------------------------------------ 3200   //------------------------------------------------
3190   // 3201   //
3191   // Measure the number of encoded characters 3202   // Measure the number of encoded characters
3192   // of output, and the number of inserted 3203   // of output, and the number of inserted
3193   // segments including internal separators. 3204   // segments including internal separators.
3194   // 3205   //
3195   3206  
HITCBC 3196   1701 std::size_t nchar = 0; 3207   1701 std::size_t nchar = 0;
HITCBC 3197   1701 std::size_t nparam = 0; 3208   1701 std::size_t nparam = 0;
HITCBC 3198   1701 if(src.measure(nchar)) 3209   1701 if(src.measure(nchar))
3199   { 3210   {
HITCBC 3200   1434 ++nchar; // for '?' or '&' 3211   1434 ++nchar; // for '?' or '&'
3201   for(;;) 3212   for(;;)
3202   { 3213   {
HITCBC 3203   1656 ++nparam; 3214   1656 ++nparam;
HITCBC 3204   1656 if(! src.measure(nchar)) 3215   1656 if(! src.measure(nchar))
HITCBC 3205   1434 break; 3216   1434 break;
HITCBC 3206   222 ++nchar; // for '&' 3217   222 ++nchar; // for '&'
3207   } 3218   }
3208   } 3219   }
3209   3220  
3210   //------------------------------------------------ 3221   //------------------------------------------------
3211   // 3222   //
3212   // Resize 3223   // Resize
3213   // 3224   //
HITCBC 3214   1696 op_t op(*this, &src.s0, &src.s1); 3225   1696 op_t op(*this, &src.s0, &src.s1);
3215   char* dest; 3226   char* dest;
3216   char const* end; 3227   char const* end;
3217   { 3228   {
HITCBC 3218   1696 auto const nremove = pos1 - pos0; 3229   1696 auto const nremove = pos1 - pos0;
3219   // check overflow 3230   // check overflow
HITCBC 3220   3005 if( nchar > nremove && 3231   3005 if( nchar > nremove &&
HITCBC 3221   1309 nchar - nremove > 3232   1309 nchar - nremove >
HITCBC 3222   1309 max_size() - size()) 3233   1309 max_size() - size())
3223   { 3234   {
3224   // too large 3235   // too large
MISUBC 3225   ✗ detail::throw_length_error(); 3236   ✗ detail::throw_length_error();
3226   } 3237   }
HITCBC 3227   1696 auto const nparam1 = 3238   1696 auto const nparam1 =
HITCBC 3228   1696 static_cast<std::ptrdiff_t>(impl_.nparam_) + 3239   1696 static_cast<std::ptrdiff_t>(impl_.nparam_) +
HITCBC 3229   1696 static_cast<std::ptrdiff_t>(nparam) - 3240   1696 static_cast<std::ptrdiff_t>(nparam) -
HITCBC 3230   1696 static_cast<std::ptrdiff_t>(it1.index) + 3241   1696 static_cast<std::ptrdiff_t>(it1.index) +
HITCBC 3231   1696 static_cast<std::ptrdiff_t>(it0.index); 3242   1696 static_cast<std::ptrdiff_t>(it0.index);
HITCBC 3232   1696 BOOST_ASSERT(nparam1 >= 0); 3243   1696 BOOST_ASSERT(nparam1 >= 0);
HITCBC 3233   1696 reserve_impl(size() + nchar - nremove, op); 3244   1696 reserve_impl(size() + nchar - nremove, op);
HITCBC 3234   1696 dest = s_ + pos0; 3245   1696 dest = s_ + pos0;
HITCBC 3235   1696 end = dest + nchar; 3246   1696 end = dest + nchar;
HITCBC 3236   1696 if(impl_.nparam_ > 0) 3247   1696 if(impl_.nparam_ > 0)
3237   { 3248   {
3238   // needed when we move 3249   // needed when we move
3239   // the beginning of the query 3250   // the beginning of the query
HITCBC 3240   1157 s_[impl_.offset(id_query)] = '&'; 3251   1157 s_[impl_.offset(id_query)] = '&';
3241   } 3252   }
HITCBC 3242   1696 op.move( 3253   1696 op.move(
HITCBC 3243   1696 dest + nchar, 3254   1696 dest + nchar,
HITCBC 3244   1696 impl_.cs_ + pos1, 3255   1696 impl_.cs_ + pos1,
HITCBC 3245   1696 size() - pos1); 3256   1696 size() - pos1);
HITCBC 3246   3392 impl_.set_size( 3257   3392 impl_.set_size(
3247   id_query, 3258   id_query,
HITCBC 3248   1696 impl_.len(id_query) + 3259   1696 impl_.len(id_query) +
3249   nchar - nremove); 3260   nchar - nremove);
HITCBC 3250   1696 impl_.nparam_ = 3261   1696 impl_.nparam_ =
HITCBC 3251   1696 detail::to_size_type(nparam1); 3262   1696 detail::to_size_type(nparam1);
HITCBC 3252   1696 if(nparam1 > 0) 3263   1696 if(nparam1 > 0)
3253   { 3264   {
3254   // needed when we erase 3265   // needed when we erase
3255   // the beginning of the query 3266   // the beginning of the query
HITCBC 3256   1593 s_[impl_.offset(id_query)] = '?'; 3267   1593 s_[impl_.offset(id_query)] = '?';
3257   } 3268   }
HITCBC 3258   1696 if(s_) 3269   1696 if(s_)
HITCBC 3259   1696 s_[size()] = '\0'; 3270   1696 s_[size()] = '\0';
3260   } 3271   }
HITCBC 3261   1696 auto const dest0 = dest; 3272   1696 auto const dest0 = dest;
3262   3273  
3263   //------------------------------------------------ 3274   //------------------------------------------------
3264   // 3275   //
3265   // Output params and internal separators: 3276   // Output params and internal separators:
3266   // 3277   //
3267   // [ '?' param ] [ '&' param ] 3278   // [ '?' param ] [ '&' param ]
3268   // 3279   //
HITCBC 3269   1696 if(nparam > 0) 3280   1696 if(nparam > 0)
3270   { 3281   {
HITCBC 3271   1434 if(it0.index == 0) 3282   1434 if(it0.index == 0)
HITCBC 3272   889 *dest++ = '?'; 3283   889 *dest++ = '?';
3273   else 3284   else
HITCBC 3274   545 *dest++ = '&'; 3285   545 *dest++ = '&';
HITCBC 3275   1434 src.rewind(); 3286   1434 src.rewind();
3276   for(;;) 3287   for(;;)
3277   { 3288   {
HITCBC 3278   1656 src.copy(dest, end); 3289   1656 src.copy(dest, end);
HITCBC 3279   1656 if(--nparam == 0) 3290   1656 if(--nparam == 0)
HITCBC 3280   1434 break; 3291   1434 break;
HITCBC 3281   222 *dest++ = '&'; 3292   222 *dest++ = '&';
3282   } 3293   }
3283   } 3294   }
3284   3295  
3285   // calc decoded size of new range; see dn0. 3296   // calc decoded size of new range; see dn0.
3286   auto dn = 3297   auto dn =
3287   static_cast<std::ptrdiff_t>( 3298   static_cast<std::ptrdiff_t>(
HITCBC 3288   1696 detail::decode_bytes_unsafe( 3299   1696 detail::decode_bytes_unsafe(
HITCBC 3289   1696 core::string_view(dest0, dest - dest0))); 3300   1696 core::string_view(dest0, dest - dest0)));
HITCBC 3290   1696 if(impl_.len(id_query) > 0) 3301   1696 if(impl_.len(id_query) > 0)
HITCBC 3291   1593 dn -= 1; 3302   1593 dn -= 1;
3292   3303  
HITCBC 3293   1696 if(dn >= dn0) 3304   1696 if(dn >= dn0)
HITCBC 3294   1314 impl_.decoded_[id_query] += 3305   1314 impl_.decoded_[id_query] +=
HITCBC 3295   1314 detail::to_size_type(dn - dn0); 3306   1314 detail::to_size_type(dn - dn0);
3296   else 3307   else
HITCBC 3297   382 impl_.decoded_[id_query] -= 3308   382 impl_.decoded_[id_query] -=
HITCBC 3298   382 detail::to_size_type(dn0 - dn); 3309   382 detail::to_size_type(dn0 - dn);
3299   3310  
3300   return detail::params_iter_impl( 3311   return detail::params_iter_impl(
HITCBC 3301   1696 impl_, 3312   1696 impl_,
HITCBC 3302   1696 pos0 - impl_.offset_[id_query], 3313   1696 pos0 - impl_.offset_[id_query],
HITCBC 3303   3392 it0.index); 3314   3392 it0.index);
HITCBC 3304   1696 } 3315   1696 }
3305   3316  
3306   //------------------------------------------------ 3317   //------------------------------------------------
3307   3318  
3308   inline 3319   inline
3309   void 3320   void
HITCBC 3310   566 url_base:: 3321   604 url_base::
3311   decoded_to_lower_impl(int id) noexcept 3322   decoded_to_lower_impl(int id) noexcept
3312   { 3323   {
HITCBC 3313   566 char* it = s_ + impl_.offset(id); 3324   604 char* it = s_ + impl_.offset(id);
HITCBC 3314   566 char const* const end = s_ + impl_.offset(id + 1); 3325   604 char const* const end = s_ + impl_.offset(id + 1);
HITCBC 3315   3562 while(it < end) 3326   3600 while(it < end)
3316   { 3327   {
HITCBC 3317   2996 if (*it != '%') 3328   2996 if (*it != '%')
3318   { 3329   {
HITCBC 3319   5962 *it = grammar::to_lower( 3330   5962 *it = grammar::to_lower(
HITCBC 3320   2981 *it); 3331   2981 *it);
HITCBC 3321   2981 ++it; 3332   2981 ++it;
HITCBC 3322   2981 continue; 3333   2981 continue;
3323   } 3334   }
HITCBC 3324   15 it += 3; 3335   15 it += 3;
3325   } 3336   }
HITCBC 3326   566 } 3337   604 }
3327   3338  
3328   inline 3339   inline
3329   void 3340   void
HITCBC 3330   220 url_base:: 3341   258 url_base::
3331   to_lower_impl(int id) noexcept 3342   to_lower_impl(int id) noexcept
3332   { 3343   {
HITCBC 3333   220 char* it = s_ + impl_.offset(id); 3344   258 char* it = s_ + impl_.offset(id);
HITCBC 3334   220 char const* const end = s_ + impl_.offset(id + 1); 3345   258 char const* const end = s_ + impl_.offset(id + 1);
HITCBC 3335   1091 while(it < end) 3346   1129 while(it < end)
3336   { 3347   {
HITCBC 3337   1742 *it = grammar::to_lower( 3348   1742 *it = grammar::to_lower(
HITCBC 3338   871 *it); 3349   871 *it);
HITCBC 3339   871 ++it; 3350   871 ++it;
3340   } 3351   }
HITCBC 3341   220 } 3352   258 }
3342   3353  
3343   } // urls 3354   } // urls
3344   } // boost 3355   } // boost
3345   3356  
3346   #endif 3357   #endif