GCC Code Coverage Report

Directory: libs/url/
File: libs/url/src/detail/pattern.cpp
Date: 2024-03-15 19:37:09
Exec Total Coverage
Lines: 356 356 100.0%
Functions: 10 10 100.0%
Branches: 188 252 74.6%
Line Branch Exec Source
1 //
2 // Copyright (c) 2022 Alan de Freitas (alandefreitas@gmail.com)
3 //
4 // Distributed under the Boost Software License, Version 1.0. (See accompanying
5 // file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
6 //
7 // Official repository: https://github.com/boostorg/url
8 //
9
10
11 #include <boost/url/detail/config.hpp>
12 #include "pattern.hpp"
13 #include "pct_format.hpp"
14 #include "boost/url/detail/replacement_field_rule.hpp"
15 #include <boost/url/grammar/alpha_chars.hpp>
16 #include <boost/url/grammar/optional_rule.hpp>
17 #include <boost/url/grammar/token_rule.hpp>
18 #include "../rfc/detail/charsets.hpp"
19 #include "../rfc/detail/host_rule.hpp"
20 #include "boost/url/rfc/detail/path_rules.hpp"
21 #include "../rfc/detail/port_rule.hpp"
22 #include "../rfc/detail/scheme_rule.hpp"
23
24 namespace boost {
25 namespace urls {
26 namespace detail {
27
28 static constexpr auto lhost_chars = host_chars + ':';
29
30 void
31 140 pattern::
32 apply(
33 url_base& u,
34 format_args const& args) const
35 {
36 // measure total
37 struct sizes
38 {
39 std::size_t scheme = 0;
40 std::size_t user = 0;
41 std::size_t pass = 0;
42 std::size_t host = 0;
43 std::size_t port = 0;
44 std::size_t path = 0;
45 std::size_t query = 0;
46 std::size_t frag = 0;
47 };
48 140 sizes n;
49
50 140 format_parse_context pctx(nullptr, nullptr, 0);
51
1/2
✓ Branch 1 taken 140 times.
✗ Branch 2 not taken.
 140 measure_context mctx(args);
52
2/2
✓ Branch 1 taken 54 times.
✓ Branch 2 taken 86 times.
 140 if (!scheme.empty())
53 {
54 54 pctx = {scheme, pctx.next_arg_id()};
55
1/2
✓ Branch 2 taken 54 times.
✗ Branch 3 not taken.
 54 n.scheme = pct_vmeasure(
56 grammar::alpha_chars, pctx, mctx);
57 54 mctx.advance_to(0);
58 }
59
2/2
✓ Branch 0 taken 47 times.
✓ Branch 1 taken 93 times.
 140 if (has_authority)
60 {
61
2/2
✓ Branch 0 taken 8 times.
✓ Branch 1 taken 39 times.
 47 if (has_user)
62 {
63 8 pctx = {user, pctx.next_arg_id()};
64
1/2
✓ Branch 1 taken 8 times.
✗ Branch 2 not taken.
 8 n.user = pct_vmeasure(
65 user_chars, pctx, mctx);
66 8 mctx.advance_to(0);
67
2/2
✓ Branch 0 taken 6 times.
✓ Branch 1 taken 2 times.
 8 if (has_pass)
68 {
69 6 pctx = {pass, pctx.next_arg_id()};
70
1/2
✓ Branch 1 taken 6 times.
✗ Branch 2 not taken.
 6 n.pass = pct_vmeasure(
71 password_chars, pctx, mctx);
72 6 mctx.advance_to(0);
73 }
74 }
75
2/2
✓ Branch 1 taken 1 times.
✓ Branch 2 taken 46 times.
 47 if (host.starts_with('['))
76 {
77
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 1 times.
 1 BOOST_ASSERT(host.ends_with(']'));
78
1/2
✓ Branch 2 taken 1 times.
✗ Branch 3 not taken.
 1 pctx = {host.substr(1, host.size() - 2), pctx.next_arg_id()};
79
1/2
✓ Branch 1 taken 1 times.
✗ Branch 2 not taken.
 1 n.host = pct_vmeasure(
80 1 lhost_chars, pctx, mctx) + 2;
81 1 mctx.advance_to(0);
82 }
83 else
84 {
85 46 pctx = {host, pctx.next_arg_id()};
86
1/2
✓ Branch 1 taken 46 times.
✗ Branch 2 not taken.
 46 n.host = pct_vmeasure(
87 host_chars, pctx, mctx);
88 46 mctx.advance_to(0);
89 }
90
2/2
✓ Branch 0 taken 13 times.
✓ Branch 1 taken 34 times.
 47 if (has_port)
91 {
92 13 pctx = {port, pctx.next_arg_id()};
93
1/2
✓ Branch 2 taken 13 times.
✗ Branch 3 not taken.
 13 n.port = pct_vmeasure(
94 grammar::digit_chars, pctx, mctx);
95 13 mctx.advance_to(0);
96 }
97 }
98
2/2
✓ Branch 1 taken 102 times.
✓ Branch 2 taken 38 times.
 140 if (!path.empty())
99 {
100 102 pctx = {path, pctx.next_arg_id()};
101
2/2
✓ Branch 1 taken 100 times.
✓ Branch 2 taken 2 times.
 102 n.path = pct_vmeasure(
102 path_chars, pctx, mctx);
103 100 mctx.advance_to(0);
104 }
105
2/2
✓ Branch 0 taken 13 times.
✓ Branch 1 taken 125 times.
 138 if (has_query)
106 {
107 13 pctx = {query, pctx.next_arg_id()};
108
1/2
✓ Branch 1 taken 13 times.
✗ Branch 2 not taken.
 13 n.query = pct_vmeasure(
109 query_chars, pctx, mctx);
110 13 mctx.advance_to(0);
111 }
112
2/2
✓ Branch 0 taken 7 times.
✓ Branch 1 taken 131 times.
 138 if (has_frag)
113 {
114 7 pctx = {frag, pctx.next_arg_id()};
115
1/2
✓ Branch 1 taken 7 times.
✗ Branch 2 not taken.
 7 n.frag = pct_vmeasure(
116 fragment_chars, pctx, mctx);
117 7 mctx.advance_to(0);
118 }
119 138 std::size_t const n_total =
120 138 n.scheme +
121 138 (n.scheme != 0) * 1 + // ":"
122 138 has_authority * 2 + // "//"
123 138 n.user +
124 138 has_pass * 1 + // ":"
125 138 n.pass +
126 138 has_user * 1 + // "@"
127 138 n.host +
128 138 has_port * 1 + // ":"
129 138 n.port +
130 138 n.path +
131 138 has_query * 1 + // "?"
132 138 n.query +
133 138 has_frag * 1 + // "#"
134 138 n.frag;
135
2/2
✓ Branch 1 taken 137 times.
✓ Branch 2 taken 1 times.
 138 u.reserve(n_total);
136
137 // Apply
138 137 pctx = {nullptr, nullptr, 0};
139 137 format_context fctx(nullptr, args);
140 274 url_base::op_t op(u);
141 using parts = parts_base;
142
2/2
✓ Branch 1 taken 53 times.
✓ Branch 2 taken 84 times.
 137 if (!scheme.empty())
143 {
144 106 auto dest = u.resize_impl(
145 parts::id_scheme,
146
1/2
✓ Branch 1 taken 53 times.
✗ Branch 2 not taken.
 53 n.scheme + 1, op);
147 53 pctx = {scheme, pctx.next_arg_id()};
148 53 fctx.advance_to(dest);
149
1/2
✓ Branch 2 taken 53 times.
✗ Branch 3 not taken.
 53 const char* dest1 = pct_vformat(
150 grammar::alpha_chars, pctx, fctx);
151 53 dest[n.scheme] = ':';
152 // validate
153
2/2
✓ Branch 3 taken 1 times.
✓ Branch 4 taken 52 times.
 53 if (!grammar::parse({dest, dest1}, scheme_rule()))
154 {
155 1 throw_invalid_argument();
156 }
157 }
158
2/2
✓ Branch 0 taken 45 times.
✓ Branch 1 taken 91 times.
 136 if (has_authority)
159 {
160
2/2
✓ Branch 0 taken 8 times.
✓ Branch 1 taken 37 times.
 45 if (has_user)
161 {
162
1/2
✓ Branch 1 taken 8 times.
✗ Branch 2 not taken.
 8 auto dest = u.set_user_impl(
163 n.user, op);
164 8 pctx = {user, pctx.next_arg_id()};
165 8 fctx.advance_to(dest);
166
1/2
✓ Branch 1 taken 8 times.
✗ Branch 2 not taken.
 8 char const* dest1 = pct_vformat(
167 user_chars, pctx, fctx);
168 8 u.impl_.decoded_[parts::id_user] =
169
1/2
✓ Branch 1 taken 8 times.
✗ Branch 2 not taken.
 8 pct_string_view(dest, dest1 - dest)
170 8 ->decoded_size();
171
2/2
✓ Branch 0 taken 6 times.
✓ Branch 1 taken 2 times.
 8 if (has_pass)
172 {
173
1/2
✓ Branch 1 taken 6 times.
✗ Branch 2 not taken.
 6 char* destp = u.set_password_impl(
174 n.pass, op);
175 6 pctx = {pass, pctx.next_arg_id()};
176 6 fctx.advance_to(destp);
177
1/2
✓ Branch 1 taken 6 times.
✗ Branch 2 not taken.
 6 dest1 = pct_vformat(
178 password_chars, pctx, fctx);
179 6 u.impl_.decoded_[parts::id_pass] =
180
1/2
✓ Branch 2 taken 6 times.
✗ Branch 3 not taken.
 6 pct_string_view({destp, dest1})
181 6 ->decoded_size() + 1;
182 }
183 }
184
1/2
✓ Branch 1 taken 45 times.
✗ Branch 2 not taken.
 45 auto dest = u.set_host_impl(
185 n.host, op);
186
2/2
✓ Branch 1 taken 1 times.
✓ Branch 2 taken 44 times.
 45 if (host.starts_with('['))
187 {
188
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 1 times.
 1 BOOST_ASSERT(host.ends_with(']'));
189
1/2
✓ Branch 2 taken 1 times.
✗ Branch 3 not taken.
 1 pctx = {host.substr(1, host.size() - 2), pctx.next_arg_id()};
190 1 *dest++ = '[';
191 1 fctx.advance_to(dest);
192 char* dest1 =
193
1/2
✓ Branch 1 taken 1 times.
✗ Branch 2 not taken.
 1 pct_vformat(lhost_chars, pctx, fctx);
194 1 *dest1++ = ']';
195 1 u.impl_.decoded_[parts::id_host] =
196
1/2
✓ Branch 1 taken 1 times.
✗ Branch 2 not taken.
 2 pct_string_view(dest - 1, dest1 - dest)
197 1 ->decoded_size();
198 }
199 else
200 {
201 44 pctx = {host, pctx.next_arg_id()};
202 44 fctx.advance_to(dest);
203 char const* dest1 =
204
1/2
✓ Branch 1 taken 44 times.
✗ Branch 2 not taken.
 44 pct_vformat(host_chars, pctx, fctx);
205 44 u.impl_.decoded_[parts::id_host] =
206
1/2
✓ Branch 1 taken 44 times.
✗ Branch 2 not taken.
 88 pct_string_view(dest, dest1 - dest)
207 44 ->decoded_size();
208 }
209 45 auto uh = u.encoded_host();
210
1/2
✓ Branch 4 taken 45 times.
✗ Branch 5 not taken.
 45 auto h = grammar::parse(uh, host_rule).value();
211 45 std::memcpy(
212 45 u.impl_.ip_addr_,
213 h.addr,
214 sizeof(u.impl_.ip_addr_));
215 45 u.impl_.host_type_ = h.host_type;
216
2/2
✓ Branch 0 taken 13 times.
✓ Branch 1 taken 32 times.
 45 if (has_port)
217 {
218
1/2
✓ Branch 1 taken 13 times.
✗ Branch 2 not taken.
 13 dest = u.set_port_impl(n.port, op);
219 13 pctx = {port, pctx.next_arg_id()};
220 13 fctx.advance_to(dest);
221
1/2
✓ Branch 2 taken 13 times.
✗ Branch 3 not taken.
 13 char const* dest1 = pct_vformat(
222 grammar::digit_chars, pctx, fctx);
223 13 u.impl_.decoded_[parts::id_port] =
224
1/2
✓ Branch 1 taken 13 times.
✗ Branch 2 not taken.
 13 pct_string_view(dest, dest1 - dest)
225 13 ->decoded_size() + 1;
226 13 core::string_view up = {dest - 1, dest1};
227
1/2
✓ Branch 3 taken 13 times.
✗ Branch 4 not taken.
 13 auto p = grammar::parse(up, detail::port_part_rule).value();
228
1/2
✓ Branch 0 taken 13 times.
✗ Branch 1 not taken.
 13 if (p.has_port)
229 13 u.impl_.port_number_ = p.port_number;
230 }
231 }
232
2/2
✓ Branch 1 taken 100 times.
✓ Branch 2 taken 36 times.
 136 if (!path.empty())
233 {
234
1/2
✓ Branch 1 taken 100 times.
✗ Branch 2 not taken.
 100 auto dest = u.resize_impl(
235 parts::id_path,
236 n.path, op);
237 100 pctx = {path, pctx.next_arg_id()};
238 100 fctx.advance_to(dest);
239
1/2
✓ Branch 1 taken 100 times.
✗ Branch 2 not taken.
 100 auto dest1 = pct_vformat(
240 path_chars, pctx, fctx);
241
1/2
✓ Branch 1 taken 100 times.
✗ Branch 2 not taken.
 100 pct_string_view npath(dest, dest1 - dest);
242 100 u.impl_.decoded_[parts::id_path] +=
243 100 npath.decoded_size();
244
1/2
✓ Branch 1 taken 100 times.
✗ Branch 2 not taken.
 100 if (!npath.empty())
245 {
246
1/2
✓ Branch 1 taken 100 times.
✗ Branch 2 not taken.
 100 u.impl_.nseg_ = std::count(
247 100 npath.begin() + 1,
248 200 npath.end(), '/') + 1;
249 }
250 // handle edge cases
251 // 1) path is first component and the
252 // first segment contains an unencoded ':'
253 // This is impossible because the template
254 // "{}" would be a host.
255
4/4
✓ Branch 2 taken 78 times.
✓ Branch 3 taken 22 times.
✓ Branch 4 taken 78 times.
✓ Branch 5 taken 22 times.
 178 if (u.scheme().empty() &&
256
1/2
✓ Branch 1 taken 78 times.
✗ Branch 2 not taken.
 78 !u.has_authority())
257 {
258 78 auto fseg = u.encoded_segments().front();
259
1/2
✓ Branch 2 taken 78 times.
✗ Branch 3 not taken.
 78 std::size_t nc = std::count(
260 78 fseg.begin(), fseg.end(), ':');
261
2/2
✓ Branch 0 taken 4 times.
✓ Branch 1 taken 74 times.
 78 if (nc)
262 {
263 4 std::size_t diff = nc * 2;
264
1/2
✓ Branch 1 taken 4 times.
✗ Branch 2 not taken.
 4 u.reserve(n_total + diff);
265 8 dest = u.resize_impl(
266 parts::id_path,
267
1/2
✓ Branch 1 taken 4 times.
✗ Branch 2 not taken.
 4 n.path + diff, op);
268 4 char* dest0 = dest + diff;
269 4 std::memmove(dest0, dest, n.path);
270
2/2
✓ Branch 0 taken 23 times.
✓ Branch 1 taken 4 times.
 27 while (dest0 != dest)
271 {
272
2/2
✓ Branch 0 taken 15 times.
✓ Branch 1 taken 8 times.
 23 if (*dest0 != ':')
273 {
274 15 *dest++ = *dest0++;
275 }
276 else
277 {
278 8 *dest++ = '%';
279 8 *dest++ = '3';
280 8 *dest++ = 'A';
281 8 dest0++;
282 }
283 }
284 }
285 }
286 // 2) url has no authority and path
287 // starts with "//"
288
4/4
✓ Branch 1 taken 86 times.
✓ Branch 2 taken 14 times.
✓ Branch 3 taken 2 times.
✓ Branch 4 taken 84 times.
 186 if (!u.has_authority() &&
289
2/2
✓ Branch 2 taken 2 times.
✓ Branch 3 taken 98 times.
 186 u.encoded_path().starts_with("//"))
290 {
291
1/2
✓ Branch 1 taken 2 times.
✗ Branch 2 not taken.
 2 u.reserve(n_total + 2);
292 4 dest = u.resize_impl(
293 parts::id_path,
294
1/2
✓ Branch 1 taken 2 times.
✗ Branch 2 not taken.
 2 n.path + 2, op);
295 2 std::memmove(dest + 2, dest, n.path);
296 2 *dest++ = '/';
297 2 *dest = '.';
298 }
299 }
300
2/2
✓ Branch 0 taken 13 times.
✓ Branch 1 taken 123 times.
 136 if (has_query)
301 {
302 26 auto dest = u.resize_impl(
303 parts::id_query,
304
1/2
✓ Branch 1 taken 13 times.
✗ Branch 2 not taken.
 13 n.query + 1, op);
305 13 *dest++ = '?';
306 13 pctx = {query, pctx.next_arg_id()};
307 13 fctx.advance_to(dest);
308
1/2
✓ Branch 1 taken 13 times.
✗ Branch 2 not taken.
 13 auto dest1 = pct_vformat(
309 query_chars, pctx, fctx);
310
1/2
✓ Branch 1 taken 13 times.
✗ Branch 2 not taken.
 13 pct_string_view nquery(dest, dest1 - dest);
311 13 u.impl_.decoded_[parts::id_query] +=
312 13 nquery.decoded_size() + 1;
313
1/2
✓ Branch 1 taken 13 times.
✗ Branch 2 not taken.
 13 if (!nquery.empty())
314 {
315
1/2
✓ Branch 2 taken 13 times.
✗ Branch 3 not taken.
 13 u.impl_.nparam_ = std::count(
316 nquery.begin(),
317 26 nquery.end(), '&') + 1;
318 }
319 }
320
2/2
✓ Branch 0 taken 7 times.
✓ Branch 1 taken 129 times.
 136 if (has_frag)
321 {
322 14 auto dest = u.resize_impl(
323 parts::id_frag,
324
1/2
✓ Branch 1 taken 7 times.
✗ Branch 2 not taken.
 7 n.frag + 1, op);
325 7 *dest++ = '#';
326 7 pctx = {frag, pctx.next_arg_id()};
327 7 fctx.advance_to(dest);
328
1/2
✓ Branch 1 taken 7 times.
✗ Branch 2 not taken.
 7 auto dest1 = pct_vformat(
329 fragment_chars, pctx, fctx);
330 7 u.impl_.decoded_[parts::id_frag] +=
331 7 make_pct_string_view(
332 7 core::string_view(dest, dest1 - dest))
333 7 ->decoded_size() + 1;
334 }
335 136 }
336
337 // This rule represents a pct-encoded string
338 // that contains an arbitrary number of
339 // replacement ids in it
340 template<class CharSet>
341 struct pct_encoded_fmt_string_rule_t
342 {
343 using value_type = pct_string_view;
344
345 constexpr
346 pct_encoded_fmt_string_rule_t(
347 CharSet const& cs) noexcept
348 : cs_(cs)
349 {
350 }
351
352 template<class CharSet_>
353 friend
354 constexpr
355 auto
356 pct_encoded_fmt_string_rule(
357 CharSet_ const& cs) noexcept ->
358 pct_encoded_fmt_string_rule_t<CharSet_>;
359
360 system::result<value_type>
361 482 parse(
362 char const*& it,
363 char const* end) const noexcept
364 {
365 482 auto const start = it;
366
2/2
✓ Branch 0 taken 1 times.
✓ Branch 1 taken 240 times.
 482 if(it == end)
367 {
368 // this might be empty
369 2 return {};
370 }
371
372 // consume some with literal rule
373 // this might be an empty literal
374 480 auto literal_rule = pct_encoded_rule(cs_);
375 480 auto rv = literal_rule.parse(it, end);
376
1/2
✓ Branch 1 taken 470 times.
✗ Branch 2 not taken.
 940 while (rv)
377 {
378 940 auto it0 = it;
379 // consume some with replacement id
380 // rule
381
2/2
✓ Branch 2 taken 240 times.
✓ Branch 3 taken 230 times.
 940 if (!replacement_field_rule.parse(it, end))
382 {
383 480 it = it0;
384 480 break;
385 }
386 460 rv = literal_rule.parse(it, end);
387 }
388
389 480 return core::string_view(start, it - start);
390 }
391
392 private:
393 CharSet cs_;
394 };
395
396 template<class CharSet>
397 constexpr
398 auto
399 pct_encoded_fmt_string_rule(
400 CharSet const& cs) noexcept ->
401 pct_encoded_fmt_string_rule_t<CharSet>
402 {
403 // If an error occurs here it means that
404 // the value of your type does not meet
405 // the requirements. Please check the
406 // documentation!
407 static_assert(
408 grammar::is_charset<CharSet>::value,
409 "CharSet requirements not met");
410
411 return pct_encoded_fmt_string_rule_t<CharSet>(cs);
412 }
413
414 // This rule represents a regular string with
415 // only chars from the specified charset and
416 // an arbitrary number of replacement ids in it
417 template<class CharSet>
418 struct fmt_token_rule_t
419 {
420 using value_type = pct_string_view;
421
422 constexpr
423 fmt_token_rule_t(
424 CharSet const& cs) noexcept
425 : cs_(cs)
426 {
427 }
428
429 template<class CharSet_>
430 friend
431 constexpr
432 auto
433 fmt_token_rule(
434 CharSet_ const& cs) noexcept ->
435 fmt_token_rule_t<CharSet_>;
436
437 system::result<value_type>
438 13 parse(
439 char const*& it,
440 char const* end) const noexcept
441 {
442 13 auto const start = it;
443
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 13 times.
 13 BOOST_ASSERT(it != end);
444 /*
445 // This should never happen because
446 // all tokens are optional and will
447 // already return `none`:
448 if(it == end)
449 {
450 BOOST_URL_RETURN_EC(
451 grammar::error::need_more);
452 }
453 */
454
455 // consume some with literal rule
456 // this might be an empty literal
457 auto partial_token_rule =
458 13 grammar::optional_rule(
459 13 grammar::token_rule(cs_));
460 26 auto rv = partial_token_rule.parse(it, end);
461
1/2
✓ Branch 1 taken 24 times.
✗ Branch 2 not taken.
 24 while (rv)
462 {
463 24 auto it0 = it;
464 // consume some with replacement id
465
2/2
✓ Branch 2 taken 13 times.
✓ Branch 3 taken 11 times.
 24 if (!replacement_field_rule.parse(it, end))
466 {
467 // no replacement and no more cs
468 // before: nothing else to consume
469 13 it = it0;
470 13 break;
471 }
472 // after {...}, consume any more chars
473 // in the charset
474 11 rv = partial_token_rule.parse(it, end);
475 }
476
477
2/2
✓ Branch 0 taken 1 times.
✓ Branch 1 taken 12 times.
 13 if(it == start)
478 {
479 // it != end but we consumed nothing
480 1 BOOST_URL_RETURN_EC(
481 grammar::error::need_more);
482 }
483
484 12 return core::string_view(start, it - start);
485 }
486
487 private:
488 CharSet cs_;
489 };
490
491 template<class CharSet>
492 constexpr
493 auto
494 fmt_token_rule(
495 CharSet const& cs) noexcept ->
496 fmt_token_rule_t<CharSet>
497 {
498 // If an error occurs here it means that
499 // the value of your type does not meet
500 // the requirements. Please check the
501 // documentation!
502 static_assert(
503 grammar::is_charset<CharSet>::value,
504 "CharSet requirements not met");
505
506 return fmt_token_rule_t<CharSet>(cs);
507 }
508
509 struct userinfo_template_rule_t
510 {
511 struct value_type
512 {
513 core::string_view user;
514 core::string_view password;
515 bool has_password = false;
516 };
517
518 auto
519 48 parse(
520 char const*& it,
521 char const* end
522 ) const noexcept ->
523 system::result<value_type>
524 {
525 static constexpr auto uchars =
526 unreserved_chars +
527 sub_delim_chars;
528 static constexpr auto pwchars =
529 uchars + ':';
530
531 48 value_type t;
532
533 // user
534 static constexpr auto user_fmt_rule =
535 pct_encoded_fmt_string_rule(uchars);
536 auto rv = grammar::parse(
537 48 it, end, user_fmt_rule);
538
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 48 times.
 48 BOOST_ASSERT(rv);
539 48 t.user = *rv;
540
541 // ':'
542
2/2
✓ Branch 0 taken 31 times.
✓ Branch 1 taken 17 times.
 48 if( it == end ||
543
2/2
✓ Branch 0 taken 15 times.
✓ Branch 1 taken 16 times.
 31 *it != ':')
544 {
545 32 t.has_password = false;
546 32 t.password = {};
547 32 return t;
548 }
549 16 ++it;
550
551 // pass
552 static constexpr auto pass_fmt_rule =
553 pct_encoded_fmt_string_rule(grammar::ref(pwchars));
554 rv = grammar::parse(
555 16 it, end, pass_fmt_rule);
556
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 16 times.
 16 BOOST_ASSERT(rv);
557 16 t.has_password = true;
558 16 t.password = *rv;
559
560 16 return t;
561 }
562 };
563
564 constexpr userinfo_template_rule_t userinfo_template_rule{};
565
566 struct host_template_rule_t
567 {
568 using value_type = core::string_view;
569
570 auto
571 49 parse(
572 char const*& it,
573 char const* end
574 ) const noexcept ->
575 system::result<value_type>
576 {
577
2/2
✓ Branch 0 taken 1 times.
✓ Branch 1 taken 48 times.
 49 if(it == end)
578 {
579 // empty host
580 1 return {};
581 }
582
583 // the host type will be ultimately
584 // validated when applying the replacement
585 // strings. Any chars allowed in hosts
586 // are allowed here.
587
2/2
✓ Branch 0 taken 46 times.
✓ Branch 1 taken 2 times.
 48 if (*it != '[')
588 {
589 // IPv4address and reg-name have the
590 // same char sets.
591 46 constexpr auto any_host_template_rule =
592 pct_encoded_fmt_string_rule(host_chars);
593 auto rv = grammar::parse(
594 46 it, end, any_host_template_rule);
595 // any_host_template_rule can always
596 // be empty, so it's never invalid
597
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 46 times.
 46 BOOST_ASSERT(rv);
598 46 return detail::to_sv(*rv);
599 }
600 // IP-literals need to be enclosed in
601 // "[]" if using ':' in the template
602 // string, because the ':' would be
603 // ambiguous with the port in fmt string.
604 // The "[]:" can be used in replacement
605 // strings without the "[]" though.
606 2 constexpr auto ip_literal_template_rule =
607 pct_encoded_fmt_string_rule(lhost_chars);
608 2 auto it0 = it;
609 auto rv = grammar::parse(
610 it, end,
611 2 grammar::optional_rule(
612 2 grammar::tuple_rule(
613 2 grammar::squelch(
614 2 grammar::delim_rule('[')),
615 ip_literal_template_rule,
616 2 grammar::squelch(
617 4 grammar::delim_rule(']')))));
618 // ip_literal_template_rule can always
619 // be empty, so it's never invalid, but
620 // the rule might fail to match the
621 // closing "]"
622
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 2 times.
 2 BOOST_ASSERT(rv);
623 2 return core::string_view{it0, it};
624 }
625 };
626
627 constexpr host_template_rule_t host_template_rule{};
628
629 struct authority_template_rule_t
630 {
631 using value_type = pattern;
632
633 system::result<value_type>
634 49 parse(
635 char const*& it,
636 char const* end
637 ) const noexcept
638 {
639 49 pattern u;
640
641 // [ userinfo "@" ]
642 {
643 auto rv = grammar::parse(
644 it, end,
645 49 grammar::optional_rule(
646 49 grammar::tuple_rule(
647 userinfo_template_rule,
648 49 grammar::squelch(
649 98 grammar::delim_rule('@')))));
650
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 49 times.
 49 BOOST_ASSERT(rv);
651
2/2
✓ Branch 2 taken 9 times.
✓ Branch 3 taken 40 times.
 49 if(rv->has_value())
652 {
653 9 auto& r = **rv;
654 9 u.has_user = true;
655 9 u.user = r.user;
656 9 u.has_pass = r.has_password;
657 9 u.pass = r.password;
658 }
659 }
660
661 // host
662 {
663 auto rv = grammar::parse(
664 it, end,
665 49 host_template_rule);
666 // host is allowed to be empty
667
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 49 times.
 49 BOOST_ASSERT(rv);
668 49 u.host = *rv;
669 }
670
671 // [ ":" port ]
672 {
673 constexpr auto port_template_rule =
674 grammar::optional_rule(
675 fmt_token_rule(grammar::digit_chars));
676 49 auto it0 = it;
677 auto rv = grammar::parse(
678 it, end,
679 49 grammar::tuple_rule(
680 49 grammar::squelch(
681 49 grammar::delim_rule(':')),
682 98 port_template_rule));
683
2/2
✓ Branch 1 taken 35 times.
✓ Branch 2 taken 14 times.
 49 if (!rv)
684 {
685 35 it = it0;
686 }
687 else
688 {
689 14 u.has_port = true;
690
2/2
✓ Branch 2 taken 12 times.
✓ Branch 3 taken 2 times.
 14 if (rv->has_value())
691 {
692 12 u.port = **rv;
693 }
694 }
695 }
696
697 49 return u;
698 }
699 };
700
701 constexpr authority_template_rule_t authority_template_rule{};
702
703 struct scheme_template_rule_t
704 {
705 using value_type = core::string_view;
706
707 system::result<value_type>
708 147 parse(
709 char const*& it,
710 char const* end) const noexcept
711 {
712 147 auto const start = it;
713
2/2
✓ Branch 0 taken 1 times.
✓ Branch 1 taken 146 times.
 147 if(it == end)
714 {
715 // scheme can't be empty
716 1 BOOST_URL_RETURN_EC(
717 grammar::error::mismatch);
718 }
719
4/4
✓ Branch 1 taken 124 times.
✓ Branch 2 taken 22 times.
✓ Branch 3 taken 20 times.
✓ Branch 4 taken 126 times.
 270 if(!grammar::alpha_chars(*it) &&
720
2/2
✓ Branch 0 taken 20 times.
✓ Branch 1 taken 104 times.
 124 *it != '{')
721 {
722 // expected alpha
723 20 BOOST_URL_RETURN_EC(
724 grammar::error::mismatch);
725 }
726
727 // it starts with replacement id or alpha char
728
2/2
✓ Branch 1 taken 104 times.
✓ Branch 2 taken 22 times.
 126 if (!grammar::alpha_chars(*it))
729 {
730
2/2
✓ Branch 2 taken 2 times.
✓ Branch 3 taken 102 times.
 104 if (!replacement_field_rule.parse(it, end))
731 {
732 // replacement_field_rule is invalid
733 2 BOOST_URL_RETURN_EC(
734 grammar::error::mismatch);
735 }
736 }
737 else
738 {
739 // skip first
740 22 ++it;
741 }
742
743 static
744 constexpr
745 grammar::lut_chars scheme_chars(
746 "0123456789" "+-."
747 "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
748 "abcdefghijklmnopqrstuvwxyz");
749
750 // non-scheme chars might be a new
751 // replacement-id or just an invalid char
752 124 it = grammar::find_if_not(
753 it, end, scheme_chars);
754
2/2
✓ Branch 0 taken 75 times.
✓ Branch 1 taken 51 times.
 126 while (it != end)
755 {
756 75 auto it0 = it;
757
2/2
✓ Branch 2 taken 73 times.
✓ Branch 3 taken 2 times.
 75 if (!replacement_field_rule.parse(it, end))
758 {
759 73 it = it0;
760 73 break;
761 }
762 2 it = grammar::find_if_not(
763 it, end, scheme_chars);
764 }
765 124 return core::string_view(start, it - start);
766 }
767 };
768
769 constexpr scheme_template_rule_t scheme_template_rule{};
770
771 // This rule should consider all url types at the
772 // same time according to the format string
773 // - relative urls with no scheme/authority
774 // - absolute urls have no fragment
775 struct pattern_rule_t
776 {
777 using value_type = pattern;
778
779 system::result<value_type>
780 147 parse(
781 char const*& it,
782 char const* const end
783 ) const noexcept
784 {
785 147 pattern u;
786
787 // optional scheme
788 {
789 147 auto it0 = it;
790 auto rv = grammar::parse(
791 it, end,
792 147 grammar::tuple_rule(
793 scheme_template_rule,
794 147 grammar::squelch(
795 147 grammar::delim_rule(':'))));
796
2/2
✓ Branch 1 taken 59 times.
✓ Branch 2 taken 88 times.
 147 if(rv)
797 59 u.scheme = *rv;
798 else
799 88 it = it0;
800 }
801
802 // hier_part (authority + path)
803 // if there are less than 2 chars left,
804 // we are parsing the path
805
2/2
✓ Branch 0 taken 4 times.
✓ Branch 1 taken 143 times.
 147 if (it == end)
806 {
807 // this is over, so we can consider
808 // that a "path-empty"
809 4 return u;
810 }
811
2/2
✓ Branch 0 taken 5 times.
✓ Branch 1 taken 138 times.
 143 if(end - it == 1)
812 {
813 // only one char left
814 // it can be a single separator "/",
815 // representing an empty absolute path,
816 // or a single-char segment
817
2/2
✓ Branch 0 taken 2 times.
✓ Branch 1 taken 3 times.
 5 if(*it == '/')
818 {
819 // path-absolute
820 2 u.path = {it, 1};
821 2 ++it;
822 2 return u;
823 }
824 // this can be a:
825 // - path-noscheme if there's no scheme, or
826 // - path-rootless with a single char, or
827 // - path-empty (and consume nothing)
828
3/4
✓ Branch 1 taken 1 times.
✓ Branch 2 taken 2 times.
✓ Branch 3 taken 3 times.
✗ Branch 4 not taken.
 4 if (!u.scheme.empty() ||
829
1/2
✓ Branch 0 taken 1 times.
✗ Branch 1 not taken.
 1 *it != ':')
830 {
831 // path-rootless with a single char
832 // this needs to be a segment because
833 // the authority needs two slashes
834 // "//"
835 // path-noscheme also matches here
836 // because we already validated the
837 // first char
838 auto rv = grammar::parse(
839 3 it, end, urls::detail::segment_rule);
840
2/2
✓ Branch 1 taken 1 times.
✓ Branch 2 taken 2 times.
 3 if(! rv)
841 1 return rv.error();
842 2 u.path = *rv;
843 }
844 2 return u;
845 }
846
847 // authority
848
2/2
✓ Branch 0 taken 62 times.
✓ Branch 1 taken 76 times.
 138 if( it[0] == '/' &&
849
2/2
✓ Branch 0 taken 49 times.
✓ Branch 1 taken 13 times.
 62 it[1] == '/')
850 {
851 // "//" always indicates authority
852 49 it += 2;
853 auto rv = grammar::parse(
854 it, end,
855 49 authority_template_rule);
856 // authority is allowed to be empty
857
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 49 times.
 49 BOOST_ASSERT(rv);
858 49 u.has_authority = true;
859 49 u.has_user = rv->has_user;
860 49 u.user = rv->user;
861 49 u.has_pass = rv->has_pass;
862 49 u.pass = rv->pass;
863 49 u.host = rv->host;
864 49 u.has_port = rv->has_port;
865 49 u.port = rv->port;
866 }
867
868 // the authority requires an absolute path
869 // or an empty path
870
2/2
✓ Branch 0 taken 111 times.
✓ Branch 1 taken 27 times.
 138 if (it == end ||
871
2/2
✓ Branch 0 taken 22 times.
✓ Branch 1 taken 89 times.
 111 (u.has_authority &&
872
2/2
✓ Branch 0 taken 8 times.
✓ Branch 1 taken 14 times.
 22 (*it != '/' &&
873
2/2
✓ Branch 0 taken 2 times.
✓ Branch 1 taken 6 times.
 8 *it != '?' &&
874
1/2
✓ Branch 0 taken 2 times.
✗ Branch 1 not taken.
 2 *it != '#')))
875 {
876 // path-empty
877 29 return u;
878 }
879
880 // path-abempty
881 // consume the whole path at once because
882 // we're going to count number of segments
883 // later after the replacements happen
884 static constexpr auto segment_fmt_rule =
885 pct_encoded_fmt_string_rule(path_chars);
886 auto rp = grammar::parse(
887 109 it, end, segment_fmt_rule);
888 // path-abempty is allowed to be empty
889
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 109 times.
 109 BOOST_ASSERT(rp);
890 109 u.path = *rp;
891
892 // [ "?" query ]
893 {
894 static constexpr auto query_fmt_rule =
895 pct_encoded_fmt_string_rule(query_chars);
896 auto rv = grammar::parse(
897 it, end,
898 109 grammar::tuple_rule(
899 109 grammar::squelch(
900 109 grammar::delim_rule('?')),
901 109 query_fmt_rule));
902 // query is allowed to be empty but
903 // delim rule is not
904
2/2
✓ Branch 1 taken 13 times.
✓ Branch 2 taken 96 times.
 109 if (rv)
905 {
906 13 u.has_query = true;
907 13 u.query = *rv;
908 }
909 }
910
911 // [ "#" fragment ]
912 {
913 static constexpr auto frag_fmt_rule =
914 pct_encoded_fmt_string_rule(fragment_chars);
915 auto rv = grammar::parse(
916 it, end,
917 109 grammar::tuple_rule(
918 109 grammar::squelch(
919 109 grammar::delim_rule('#')),
920 109 frag_fmt_rule));
921 // frag is allowed to be empty but
922 // delim rule is not
923
2/2
✓ Branch 1 taken 7 times.
✓ Branch 2 taken 102 times.
 109 if (rv)
924 {
925 7 u.has_frag = true;
926 7 u.frag = *rv;
927 }
928 }
929
930 109 return u;
931 }
932 };
933
934 constexpr pattern_rule_t pattern_rule{};
935
936 system::result<pattern>
937 147 parse_pattern(
938 core::string_view s)
939 {
940 return grammar::parse(
941 147 s, pattern_rule);
942 }
943
944 } // detail
945 } // urls
946 } // boost
947
948