GCC Code Coverage Report

Directory: libs/url/
File: libs/url/src/detail/pattern.cpp
Date: 2024-01-19 15:42:54
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 #ifndef BOOST_URL_DETAIL_IMPL_PATTERN_IPP
11 #define BOOST_URL_DETAIL_IMPL_PATTERN_IPP
12
13 #include <boost/url/detail/config.hpp>
14 #include "pattern.hpp"
15 #include "pct_format.hpp"
16 #include "boost/url/detail/replacement_field_rule.hpp"
17 #include <boost/url/grammar/alpha_chars.hpp>
18 #include <boost/url/grammar/optional_rule.hpp>
19 #include <boost/url/grammar/token_rule.hpp>
20 #include "../rfc/detail/charsets.hpp"
21 #include "../rfc/detail/host_rule.hpp"
22 #include "boost/url/rfc/detail/path_rules.hpp"
23 #include "../rfc/detail/port_rule.hpp"
24 #include "../rfc/detail/scheme_rule.hpp"
25
26 namespace boost {
27 namespace urls {
28 namespace detail {
29
30 static constexpr auto lhost_chars = host_chars + ':';
31
32 void
33 140 pattern::
34 apply(
35 url_base& u,
36 format_args const& args) const
37 {
38 // measure total
39 struct sizes
40 {
41 std::size_t scheme = 0;
42 std::size_t user = 0;
43 std::size_t pass = 0;
44 std::size_t host = 0;
45 std::size_t port = 0;
46 std::size_t path = 0;
47 std::size_t query = 0;
48 std::size_t frag = 0;
49 };
50 140 sizes n;
51
52 140 format_parse_context pctx(nullptr, nullptr, 0);
53
1/2
✓ Branch 1 taken 140 times.
✗ Branch 2 not taken.
 140 measure_context mctx(args);
54
2/2
✓ Branch 1 taken 54 times.
✓ Branch 2 taken 86 times.
 140 if (!scheme.empty())
55 {
56 54 pctx = {scheme, pctx.next_arg_id()};
57
1/2
✓ Branch 2 taken 54 times.
✗ Branch 3 not taken.
 54 n.scheme = pct_vmeasure(
58 grammar::alpha_chars, pctx, mctx);
59 54 mctx.advance_to(0);
60 }
61
2/2
✓ Branch 0 taken 47 times.
✓ Branch 1 taken 93 times.
 140 if (has_authority)
62 {
63
2/2
✓ Branch 0 taken 8 times.
✓ Branch 1 taken 39 times.
 47 if (has_user)
64 {
65 8 pctx = {user, pctx.next_arg_id()};
66
1/2
✓ Branch 1 taken 8 times.
✗ Branch 2 not taken.
 8 n.user = pct_vmeasure(
67 user_chars, pctx, mctx);
68 8 mctx.advance_to(0);
69
2/2
✓ Branch 0 taken 6 times.
✓ Branch 1 taken 2 times.
 8 if (has_pass)
70 {
71 6 pctx = {pass, pctx.next_arg_id()};
72
1/2
✓ Branch 1 taken 6 times.
✗ Branch 2 not taken.
 6 n.pass = pct_vmeasure(
73 password_chars, pctx, mctx);
74 6 mctx.advance_to(0);
75 }
76 }
77
2/2
✓ Branch 1 taken 1 times.
✓ Branch 2 taken 46 times.
 47 if (host.starts_with('['))
78 {
79
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 1 times.
 1 BOOST_ASSERT(host.ends_with(']'));
80
1/2
✓ Branch 2 taken 1 times.
✗ Branch 3 not taken.
 1 pctx = {host.substr(1, host.size() - 2), pctx.next_arg_id()};
81
1/2
✓ Branch 1 taken 1 times.
✗ Branch 2 not taken.
 1 n.host = pct_vmeasure(
82 1 lhost_chars, pctx, mctx) + 2;
83 1 mctx.advance_to(0);
84 }
85 else
86 {
87 46 pctx = {host, pctx.next_arg_id()};
88
1/2
✓ Branch 1 taken 46 times.
✗ Branch 2 not taken.
 46 n.host = pct_vmeasure(
89 host_chars, pctx, mctx);
90 46 mctx.advance_to(0);
91 }
92
2/2
✓ Branch 0 taken 13 times.
✓ Branch 1 taken 34 times.
 47 if (has_port)
93 {
94 13 pctx = {port, pctx.next_arg_id()};
95
1/2
✓ Branch 2 taken 13 times.
✗ Branch 3 not taken.
 13 n.port = pct_vmeasure(
96 grammar::digit_chars, pctx, mctx);
97 13 mctx.advance_to(0);
98 }
99 }
100
2/2
✓ Branch 1 taken 102 times.
✓ Branch 2 taken 38 times.
 140 if (!path.empty())
101 {
102 102 pctx = {path, pctx.next_arg_id()};
103
2/2
✓ Branch 1 taken 100 times.
✓ Branch 2 taken 2 times.
 102 n.path = pct_vmeasure(
104 path_chars, pctx, mctx);
105 100 mctx.advance_to(0);
106 }
107
2/2
✓ Branch 0 taken 13 times.
✓ Branch 1 taken 125 times.
 138 if (has_query)
108 {
109 13 pctx = {query, pctx.next_arg_id()};
110
1/2
✓ Branch 1 taken 13 times.
✗ Branch 2 not taken.
 13 n.query = pct_vmeasure(
111 query_chars, pctx, mctx);
112 13 mctx.advance_to(0);
113 }
114
2/2
✓ Branch 0 taken 7 times.
✓ Branch 1 taken 131 times.
 138 if (has_frag)
115 {
116 7 pctx = {frag, pctx.next_arg_id()};
117
1/2
✓ Branch 1 taken 7 times.
✗ Branch 2 not taken.
 7 n.frag = pct_vmeasure(
118 fragment_chars, pctx, mctx);
119 7 mctx.advance_to(0);
120 }
121 138 std::size_t const n_total =
122 138 n.scheme +
123 138 (n.scheme != 0) * 1 + // ":"
124 138 has_authority * 2 + // "//"
125 138 n.user +
126 138 has_pass * 1 + // ":"
127 138 n.pass +
128 138 has_user * 1 + // "@"
129 138 n.host +
130 138 has_port * 1 + // ":"
131 138 n.port +
132 138 n.path +
133 138 has_query * 1 + // "?"
134 138 n.query +
135 138 has_frag * 1 + // "#"
136 138 n.frag;
137
2/2
✓ Branch 1 taken 137 times.
✓ Branch 2 taken 1 times.
 138 u.reserve(n_total);
138
139 // Apply
140 137 pctx = {nullptr, nullptr, 0};
141 137 format_context fctx(nullptr, args);
142 274 url_base::op_t op(u);
143 using parts = parts_base;
144
2/2
✓ Branch 1 taken 53 times.
✓ Branch 2 taken 84 times.
 137 if (!scheme.empty())
145 {
146 106 auto dest = u.resize_impl(
147 parts::id_scheme,
148
1/2
✓ Branch 1 taken 53 times.
✗ Branch 2 not taken.
 53 n.scheme + 1, op);
149 53 pctx = {scheme, pctx.next_arg_id()};
150 53 fctx.advance_to(dest);
151
1/2
✓ Branch 2 taken 53 times.
✗ Branch 3 not taken.
 53 const char* dest1 = pct_vformat(
152 grammar::alpha_chars, pctx, fctx);
153 53 dest[n.scheme] = ':';
154 // validate
155
2/2
✓ Branch 3 taken 1 times.
✓ Branch 4 taken 52 times.
 53 if (!grammar::parse({dest, dest1}, scheme_rule()))
156 {
157 1 throw_invalid_argument();
158 }
159 }
160
2/2
✓ Branch 0 taken 45 times.
✓ Branch 1 taken 91 times.
 136 if (has_authority)
161 {
162
2/2
✓ Branch 0 taken 8 times.
✓ Branch 1 taken 37 times.
 45 if (has_user)
163 {
164
1/2
✓ Branch 1 taken 8 times.
✗ Branch 2 not taken.
 8 auto dest = u.set_user_impl(
165 n.user, op);
166 8 pctx = {user, pctx.next_arg_id()};
167 8 fctx.advance_to(dest);
168
1/2
✓ Branch 1 taken 8 times.
✗ Branch 2 not taken.
 8 char const* dest1 = pct_vformat(
169 user_chars, pctx, fctx);
170 8 u.impl_.decoded_[parts::id_user] =
171
1/2
✓ Branch 1 taken 8 times.
✗ Branch 2 not taken.
 8 pct_string_view(dest, dest1 - dest)
172 8 ->decoded_size();
173
2/2
✓ Branch 0 taken 6 times.
✓ Branch 1 taken 2 times.
 8 if (has_pass)
174 {
175
1/2
✓ Branch 1 taken 6 times.
✗ Branch 2 not taken.
 6 char* destp = u.set_password_impl(
176 n.pass, op);
177 6 pctx = {pass, pctx.next_arg_id()};
178 6 fctx.advance_to(destp);
179
1/2
✓ Branch 1 taken 6 times.
✗ Branch 2 not taken.
 6 dest1 = pct_vformat(
180 password_chars, pctx, fctx);
181 6 u.impl_.decoded_[parts::id_pass] =
182
1/2
✓ Branch 2 taken 6 times.
✗ Branch 3 not taken.
 6 pct_string_view({destp, dest1})
183 6 ->decoded_size() + 1;
184 }
185 }
186
1/2
✓ Branch 1 taken 45 times.
✗ Branch 2 not taken.
 45 auto dest = u.set_host_impl(
187 n.host, op);
188
2/2
✓ Branch 1 taken 1 times.
✓ Branch 2 taken 44 times.
 45 if (host.starts_with('['))
189 {
190
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 1 times.
 1 BOOST_ASSERT(host.ends_with(']'));
191
1/2
✓ Branch 2 taken 1 times.
✗ Branch 3 not taken.
 1 pctx = {host.substr(1, host.size() - 2), pctx.next_arg_id()};
192 1 *dest++ = '[';
193 1 fctx.advance_to(dest);
194 char* dest1 =
195
1/2
✓ Branch 1 taken 1 times.
✗ Branch 2 not taken.
 1 pct_vformat(lhost_chars, pctx, fctx);
196 1 *dest1++ = ']';
197 1 u.impl_.decoded_[parts::id_host] =
198
1/2
✓ Branch 1 taken 1 times.
✗ Branch 2 not taken.
 2 pct_string_view(dest - 1, dest1 - dest)
199 1 ->decoded_size();
200 }
201 else
202 {
203 44 pctx = {host, pctx.next_arg_id()};
204 44 fctx.advance_to(dest);
205 char const* dest1 =
206
1/2
✓ Branch 1 taken 44 times.
✗ Branch 2 not taken.
 44 pct_vformat(host_chars, pctx, fctx);
207 44 u.impl_.decoded_[parts::id_host] =
208
1/2
✓ Branch 1 taken 44 times.
✗ Branch 2 not taken.
 88 pct_string_view(dest, dest1 - dest)
209 44 ->decoded_size();
210 }
211 45 auto uh = u.encoded_host();
212
1/2
✓ Branch 4 taken 45 times.
✗ Branch 5 not taken.
 45 auto h = grammar::parse(uh, host_rule).value();
213 45 std::memcpy(
214 45 u.impl_.ip_addr_,
215 h.addr,
216 sizeof(u.impl_.ip_addr_));
217 45 u.impl_.host_type_ = h.host_type;
218
2/2
✓ Branch 0 taken 13 times.
✓ Branch 1 taken 32 times.
 45 if (has_port)
219 {
220
1/2
✓ Branch 1 taken 13 times.
✗ Branch 2 not taken.
 13 dest = u.set_port_impl(n.port, op);
221 13 pctx = {port, pctx.next_arg_id()};
222 13 fctx.advance_to(dest);
223
1/2
✓ Branch 2 taken 13 times.
✗ Branch 3 not taken.
 13 char const* dest1 = pct_vformat(
224 grammar::digit_chars, pctx, fctx);
225 13 u.impl_.decoded_[parts::id_port] =
226
1/2
✓ Branch 1 taken 13 times.
✗ Branch 2 not taken.
 13 pct_string_view(dest, dest1 - dest)
227 13 ->decoded_size() + 1;
228 13 core::string_view up = {dest - 1, dest1};
229
1/2
✓ Branch 3 taken 13 times.
✗ Branch 4 not taken.
 13 auto p = grammar::parse(up, detail::port_part_rule).value();
230
1/2
✓ Branch 0 taken 13 times.
✗ Branch 1 not taken.
 13 if (p.has_port)
231 13 u.impl_.port_number_ = p.port_number;
232 }
233 }
234
2/2
✓ Branch 1 taken 100 times.
✓ Branch 2 taken 36 times.
 136 if (!path.empty())
235 {
236
1/2
✓ Branch 1 taken 100 times.
✗ Branch 2 not taken.
 100 auto dest = u.resize_impl(
237 parts::id_path,
238 n.path, op);
239 100 pctx = {path, pctx.next_arg_id()};
240 100 fctx.advance_to(dest);
241
1/2
✓ Branch 1 taken 100 times.
✗ Branch 2 not taken.
 100 auto dest1 = pct_vformat(
242 path_chars, pctx, fctx);
243
1/2
✓ Branch 1 taken 100 times.
✗ Branch 2 not taken.
 100 pct_string_view npath(dest, dest1 - dest);
244 100 u.impl_.decoded_[parts::id_path] +=
245 100 npath.decoded_size();
246
1/2
✓ Branch 1 taken 100 times.
✗ Branch 2 not taken.
 100 if (!npath.empty())
247 {
248
1/2
✓ Branch 1 taken 100 times.
✗ Branch 2 not taken.
 100 u.impl_.nseg_ = std::count(
249 100 npath.begin() + 1,
250 200 npath.end(), '/') + 1;
251 }
252 // handle edge cases
253 // 1) path is first component and the
254 // first segment contains an unencoded ':'
255 // This is impossible because the template
256 // "{}" would be a host.
257
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() &&
258
1/2
✓ Branch 1 taken 78 times.
✗ Branch 2 not taken.
 78 !u.has_authority())
259 {
260 78 auto fseg = u.encoded_segments().front();
261
1/2
✓ Branch 2 taken 78 times.
✗ Branch 3 not taken.
 78 std::size_t nc = std::count(
262 78 fseg.begin(), fseg.end(), ':');
263
2/2
✓ Branch 0 taken 4 times.
✓ Branch 1 taken 74 times.
 78 if (nc)
264 {
265 4 std::size_t diff = nc * 2;
266
1/2
✓ Branch 1 taken 4 times.
✗ Branch 2 not taken.
 4 u.reserve(n_total + diff);
267 8 dest = u.resize_impl(
268 parts::id_path,
269
1/2
✓ Branch 1 taken 4 times.
✗ Branch 2 not taken.
 4 n.path + diff, op);
270 4 char* dest0 = dest + diff;
271 4 std::memmove(dest0, dest, n.path);
272
2/2
✓ Branch 0 taken 23 times.
✓ Branch 1 taken 4 times.
 27 while (dest0 != dest)
273 {
274
2/2
✓ Branch 0 taken 15 times.
✓ Branch 1 taken 8 times.
 23 if (*dest0 != ':')
275 {
276 15 *dest++ = *dest0++;
277 }
278 else
279 {
280 8 *dest++ = '%';
281 8 *dest++ = '3';
282 8 *dest++ = 'A';
283 8 dest0++;
284 }
285 }
286 }
287 }
288 // 2) url has no authority and path
289 // starts with "//"
290
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() &&
291
2/2
✓ Branch 2 taken 2 times.
✓ Branch 3 taken 98 times.
 186 u.encoded_path().starts_with("//"))
292 {
293
1/2
✓ Branch 1 taken 2 times.
✗ Branch 2 not taken.
 2 u.reserve(n_total + 2);
294 4 dest = u.resize_impl(
295 parts::id_path,
296
1/2
✓ Branch 1 taken 2 times.
✗ Branch 2 not taken.
 2 n.path + 2, op);
297 2 std::memmove(dest + 2, dest, n.path);
298 2 *dest++ = '/';
299 2 *dest = '.';
300 }
301 }
302
2/2
✓ Branch 0 taken 13 times.
✓ Branch 1 taken 123 times.
 136 if (has_query)
303 {
304 26 auto dest = u.resize_impl(
305 parts::id_query,
306
1/2
✓ Branch 1 taken 13 times.
✗ Branch 2 not taken.
 13 n.query + 1, op);
307 13 *dest++ = '?';
308 13 pctx = {query, pctx.next_arg_id()};
309 13 fctx.advance_to(dest);
310
1/2
✓ Branch 1 taken 13 times.
✗ Branch 2 not taken.
 13 auto dest1 = pct_vformat(
311 query_chars, pctx, fctx);
312
1/2
✓ Branch 1 taken 13 times.
✗ Branch 2 not taken.
 13 pct_string_view nquery(dest, dest1 - dest);
313 13 u.impl_.decoded_[parts::id_query] +=
314 13 nquery.decoded_size() + 1;
315
1/2
✓ Branch 1 taken 13 times.
✗ Branch 2 not taken.
 13 if (!nquery.empty())
316 {
317
1/2
✓ Branch 2 taken 13 times.
✗ Branch 3 not taken.
 13 u.impl_.nparam_ = std::count(
318 nquery.begin(),
319 26 nquery.end(), '&') + 1;
320 }
321 }
322
2/2
✓ Branch 0 taken 7 times.
✓ Branch 1 taken 129 times.
 136 if (has_frag)
323 {
324 14 auto dest = u.resize_impl(
325 parts::id_frag,
326
1/2
✓ Branch 1 taken 7 times.
✗ Branch 2 not taken.
 7 n.frag + 1, op);
327 7 *dest++ = '#';
328 7 pctx = {frag, pctx.next_arg_id()};
329 7 fctx.advance_to(dest);
330
1/2
✓ Branch 1 taken 7 times.
✗ Branch 2 not taken.
 7 auto dest1 = pct_vformat(
331 fragment_chars, pctx, fctx);
332 7 u.impl_.decoded_[parts::id_frag] +=
333 7 make_pct_string_view(
334 7 core::string_view(dest, dest1 - dest))
335 7 ->decoded_size() + 1;
336 }
337 136 }
338
339 // This rule represents a pct-encoded string
340 // that contains an arbitrary number of
341 // replacement ids in it
342 template<class CharSet>
343 struct pct_encoded_fmt_string_rule_t
344 {
345 using value_type = pct_string_view;
346
347 constexpr
348 pct_encoded_fmt_string_rule_t(
349 CharSet const& cs) noexcept
350 : cs_(cs)
351 {
352 }
353
354 template<class CharSet_>
355 friend
356 constexpr
357 auto
358 pct_encoded_fmt_string_rule(
359 CharSet_ const& cs) noexcept ->
360 pct_encoded_fmt_string_rule_t<CharSet_>;
361
362 system::result<value_type>
363 482 parse(
364 char const*& it,
365 char const* end) const noexcept
366 {
367 482 auto const start = it;
368
2/2
✓ Branch 0 taken 1 times.
✓ Branch 1 taken 240 times.
 482 if(it == end)
369 {
370 // this might be empty
371 2 return {};
372 }
373
374 // consume some with literal rule
375 // this might be an empty literal
376 480 auto literal_rule = pct_encoded_rule(cs_);
377 480 auto rv = literal_rule.parse(it, end);
378
1/2
✓ Branch 1 taken 470 times.
✗ Branch 2 not taken.
 940 while (rv)
379 {
380 940 auto it0 = it;
381 // consume some with replacement id
382 // rule
383
2/2
✓ Branch 2 taken 240 times.
✓ Branch 3 taken 230 times.
 940 if (!replacement_field_rule.parse(it, end))
384 {
385 480 it = it0;
386 480 break;
387 }
388 460 rv = literal_rule.parse(it, end);
389 }
390
391 480 return core::string_view(start, it - start);
392 }
393
394 private:
395 CharSet cs_;
396 };
397
398 template<class CharSet>
399 constexpr
400 auto
401 pct_encoded_fmt_string_rule(
402 CharSet const& cs) noexcept ->
403 pct_encoded_fmt_string_rule_t<CharSet>
404 {
405 // If an error occurs here it means that
406 // the value of your type does not meet
407 // the requirements. Please check the
408 // documentation!
409 static_assert(
410 grammar::is_charset<CharSet>::value,
411 "CharSet requirements not met");
412
413 return pct_encoded_fmt_string_rule_t<CharSet>(cs);
414 }
415
416 // This rule represents a regular string with
417 // only chars from the specified charset and
418 // an arbitrary number of replacement ids in it
419 template<class CharSet>
420 struct fmt_token_rule_t
421 {
422 using value_type = pct_string_view;
423
424 constexpr
425 fmt_token_rule_t(
426 CharSet const& cs) noexcept
427 : cs_(cs)
428 {
429 }
430
431 template<class CharSet_>
432 friend
433 constexpr
434 auto
435 fmt_token_rule(
436 CharSet_ const& cs) noexcept ->
437 fmt_token_rule_t<CharSet_>;
438
439 system::result<value_type>
440 13 parse(
441 char const*& it,
442 char const* end) const noexcept
443 {
444 13 auto const start = it;
445
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 13 times.
 13 BOOST_ASSERT(it != end);
446 /*
447 // This should never happen because
448 // all tokens are optional and will
449 // already return `none`:
450 if(it == end)
451 {
452 BOOST_URL_RETURN_EC(
453 grammar::error::need_more);
454 }
455 */
456
457 // consume some with literal rule
458 // this might be an empty literal
459 auto partial_token_rule =
460 13 grammar::optional_rule(
461 13 grammar::token_rule(cs_));
462 26 auto rv = partial_token_rule.parse(it, end);
463
1/2
✓ Branch 1 taken 24 times.
✗ Branch 2 not taken.
 24 while (rv)
464 {
465 24 auto it0 = it;
466 // consume some with replacement id
467
2/2
✓ Branch 2 taken 13 times.
✓ Branch 3 taken 11 times.
 24 if (!replacement_field_rule.parse(it, end))
468 {
469 // no replacement and no more cs
470 // before: nothing else to consume
471 13 it = it0;
472 13 break;
473 }
474 // after {...}, consume any more chars
475 // in the charset
476 11 rv = partial_token_rule.parse(it, end);
477 }
478
479
2/2
✓ Branch 0 taken 1 times.
✓ Branch 1 taken 12 times.
 13 if(it == start)
480 {
481 // it != end but we consumed nothing
482 1 BOOST_URL_RETURN_EC(
483 grammar::error::need_more);
484 }
485
486 12 return core::string_view(start, it - start);
487 }
488
489 private:
490 CharSet cs_;
491 };
492
493 template<class CharSet>
494 constexpr
495 auto
496 fmt_token_rule(
497 CharSet const& cs) noexcept ->
498 fmt_token_rule_t<CharSet>
499 {
500 // If an error occurs here it means that
501 // the value of your type does not meet
502 // the requirements. Please check the
503 // documentation!
504 static_assert(
505 grammar::is_charset<CharSet>::value,
506 "CharSet requirements not met");
507
508 return fmt_token_rule_t<CharSet>(cs);
509 }
510
511 struct userinfo_template_rule_t
512 {
513 struct value_type
514 {
515 core::string_view user;
516 core::string_view password;
517 bool has_password = false;
518 };
519
520 auto
521 48 parse(
522 char const*& it,
523 char const* end
524 ) const noexcept ->
525 system::result<value_type>
526 {
527 static constexpr auto uchars =
528 unreserved_chars +
529 sub_delim_chars;
530 static constexpr auto pwchars =
531 uchars + ':';
532
533 48 value_type t;
534
535 // user
536 static constexpr auto user_fmt_rule =
537 pct_encoded_fmt_string_rule(uchars);
538 auto rv = grammar::parse(
539 48 it, end, user_fmt_rule);
540
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 48 times.
 48 BOOST_ASSERT(rv);
541 48 t.user = *rv;
542
543 // ':'
544
2/2
✓ Branch 0 taken 31 times.
✓ Branch 1 taken 17 times.
 48 if( it == end ||
545
2/2
✓ Branch 0 taken 15 times.
✓ Branch 1 taken 16 times.
 31 *it != ':')
546 {
547 32 t.has_password = false;
548 32 t.password = {};
549 32 return t;
550 }
551 16 ++it;
552
553 // pass
554 static constexpr auto pass_fmt_rule =
555 pct_encoded_fmt_string_rule(grammar::ref(pwchars));
556 rv = grammar::parse(
557 16 it, end, pass_fmt_rule);
558
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 16 times.
 16 BOOST_ASSERT(rv);
559 16 t.has_password = true;
560 16 t.password = *rv;
561
562 16 return t;
563 }
564 };
565
566 constexpr userinfo_template_rule_t userinfo_template_rule{};
567
568 struct host_template_rule_t
569 {
570 using value_type = core::string_view;
571
572 auto
573 49 parse(
574 char const*& it,
575 char const* end
576 ) const noexcept ->
577 system::result<value_type>
578 {
579
2/2
✓ Branch 0 taken 1 times.
✓ Branch 1 taken 48 times.
 49 if(it == end)
580 {
581 // empty host
582 1 return {};
583 }
584
585 // the host type will be ultimately
586 // validated when applying the replacement
587 // strings. Any chars allowed in hosts
588 // are allowed here.
589
2/2
✓ Branch 0 taken 46 times.
✓ Branch 1 taken 2 times.
 48 if (*it != '[')
590 {
591 // IPv4address and reg-name have the
592 // same char sets.
593 46 constexpr auto any_host_template_rule =
594 pct_encoded_fmt_string_rule(host_chars);
595 auto rv = grammar::parse(
596 46 it, end, any_host_template_rule);
597 // any_host_template_rule can always
598 // be empty, so it's never invalid
599
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 46 times.
 46 BOOST_ASSERT(rv);
600 46 return detail::to_sv(*rv);
601 }
602 // IP-literals need to be enclosed in
603 // "[]" if using ':' in the template
604 // string, because the ':' would be
605 // ambiguous with the port in fmt string.
606 // The "[]:" can be used in replacement
607 // strings without the "[]" though.
608 2 constexpr auto ip_literal_template_rule =
609 pct_encoded_fmt_string_rule(lhost_chars);
610 2 auto it0 = it;
611 auto rv = grammar::parse(
612 it, end,
613 2 grammar::optional_rule(
614 2 grammar::tuple_rule(
615 2 grammar::squelch(
616 2 grammar::delim_rule('[')),
617 ip_literal_template_rule,
618 2 grammar::squelch(
619 4 grammar::delim_rule(']')))));
620 // ip_literal_template_rule can always
621 // be empty, so it's never invalid, but
622 // the rule might fail to match the
623 // closing "]"
624
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 2 times.
 2 BOOST_ASSERT(rv);
625 2 return core::string_view{it0, it};
626 }
627 };
628
629 constexpr host_template_rule_t host_template_rule{};
630
631 struct authority_template_rule_t
632 {
633 using value_type = pattern;
634
635 system::result<value_type>
636 49 parse(
637 char const*& it,
638 char const* end
639 ) const noexcept
640 {
641 49 pattern u;
642
643 // [ userinfo "@" ]
644 {
645 auto rv = grammar::parse(
646 it, end,
647 49 grammar::optional_rule(
648 49 grammar::tuple_rule(
649 userinfo_template_rule,
650 49 grammar::squelch(
651 98 grammar::delim_rule('@')))));
652
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 49 times.
 49 BOOST_ASSERT(rv);
653
2/2
✓ Branch 2 taken 9 times.
✓ Branch 3 taken 40 times.
 49 if(rv->has_value())
654 {
655 9 auto& r = **rv;
656 9 u.has_user = true;
657 9 u.user = r.user;
658 9 u.has_pass = r.has_password;
659 9 u.pass = r.password;
660 }
661 }
662
663 // host
664 {
665 auto rv = grammar::parse(
666 it, end,
667 49 host_template_rule);
668 // host is allowed to be empty
669
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 49 times.
 49 BOOST_ASSERT(rv);
670 49 u.host = *rv;
671 }
672
673 // [ ":" port ]
674 {
675 constexpr auto port_template_rule =
676 grammar::optional_rule(
677 fmt_token_rule(grammar::digit_chars));
678 49 auto it0 = it;
679 auto rv = grammar::parse(
680 it, end,
681 49 grammar::tuple_rule(
682 49 grammar::squelch(
683 49 grammar::delim_rule(':')),
684 98 port_template_rule));
685
2/2
✓ Branch 1 taken 35 times.
✓ Branch 2 taken 14 times.
 49 if (!rv)
686 {
687 35 it = it0;
688 }
689 else
690 {
691 14 u.has_port = true;
692
2/2
✓ Branch 2 taken 12 times.
✓ Branch 3 taken 2 times.
 14 if (rv->has_value())
693 {
694 12 u.port = **rv;
695 }
696 }
697 }
698
699 49 return u;
700 }
701 };
702
703 constexpr authority_template_rule_t authority_template_rule{};
704
705 struct scheme_template_rule_t
706 {
707 using value_type = core::string_view;
708
709 system::result<value_type>
710 147 parse(
711 char const*& it,
712 char const* end) const noexcept
713 {
714 147 auto const start = it;
715
2/2
✓ Branch 0 taken 1 times.
✓ Branch 1 taken 146 times.
 147 if(it == end)
716 {
717 // scheme can't be empty
718 1 BOOST_URL_RETURN_EC(
719 grammar::error::mismatch);
720 }
721
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) &&
722
2/2
✓ Branch 0 taken 20 times.
✓ Branch 1 taken 104 times.
 124 *it != '{')
723 {
724 // expected alpha
725 20 BOOST_URL_RETURN_EC(
726 grammar::error::mismatch);
727 }
728
729 // it starts with replacement id or alpha char
730
2/2
✓ Branch 1 taken 104 times.
✓ Branch 2 taken 22 times.
 126 if (!grammar::alpha_chars(*it))
731 {
732
2/2
✓ Branch 2 taken 2 times.
✓ Branch 3 taken 102 times.
 104 if (!replacement_field_rule.parse(it, end))
733 {
734 // replacement_field_rule is invalid
735 2 BOOST_URL_RETURN_EC(
736 grammar::error::mismatch);
737 }
738 }
739 else
740 {
741 // skip first
742 22 ++it;
743 }
744
745 static
746 constexpr
747 grammar::lut_chars scheme_chars(
748 "0123456789" "+-."
749 "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
750 "abcdefghijklmnopqrstuvwxyz");
751
752 // non-scheme chars might be a new
753 // replacement-id or just an invalid char
754 124 it = grammar::find_if_not(
755 it, end, scheme_chars);
756
2/2
✓ Branch 0 taken 75 times.
✓ Branch 1 taken 51 times.
 126 while (it != end)
757 {
758 75 auto it0 = it;
759
2/2
✓ Branch 2 taken 73 times.
✓ Branch 3 taken 2 times.
 75 if (!replacement_field_rule.parse(it, end))
760 {
761 73 it = it0;
762 73 break;
763 }
764 2 it = grammar::find_if_not(
765 it, end, scheme_chars);
766 }
767 124 return core::string_view(start, it - start);
768 }
769 };
770
771 constexpr scheme_template_rule_t scheme_template_rule{};
772
773 // This rule should consider all url types at the
774 // same time according to the format string
775 // - relative urls with no scheme/authority
776 // - absolute urls have no fragment
777 struct pattern_rule_t
778 {
779 using value_type = pattern;
780
781 system::result<value_type>
782 147 parse(
783 char const*& it,
784 char const* const end
785 ) const noexcept
786 {
787 147 pattern u;
788
789 // optional scheme
790 {
791 147 auto it0 = it;
792 auto rv = grammar::parse(
793 it, end,
794 147 grammar::tuple_rule(
795 scheme_template_rule,
796 147 grammar::squelch(
797 147 grammar::delim_rule(':'))));
798
2/2
✓ Branch 1 taken 59 times.
✓ Branch 2 taken 88 times.
 147 if(rv)
799 59 u.scheme = *rv;
800 else
801 88 it = it0;
802 }
803
804 // hier_part (authority + path)
805 // if there are less than 2 chars left,
806 // we are parsing the path
807
2/2
✓ Branch 0 taken 4 times.
✓ Branch 1 taken 143 times.
 147 if (it == end)
808 {
809 // this is over, so we can consider
810 // that a "path-empty"
811 4 return u;
812 }
813
2/2
✓ Branch 0 taken 5 times.
✓ Branch 1 taken 138 times.
 143 if(end - it == 1)
814 {
815 // only one char left
816 // it can be a single separator "/",
817 // representing an empty absolute path,
818 // or a single-char segment
819
2/2
✓ Branch 0 taken 2 times.
✓ Branch 1 taken 3 times.
 5 if(*it == '/')
820 {
821 // path-absolute
822 2 u.path = {it, 1};
823 2 ++it;
824 2 return u;
825 }
826 // this can be a:
827 // - path-noscheme if there's no scheme, or
828 // - path-rootless with a single char, or
829 // - path-empty (and consume nothing)
830
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() ||
831
1/2
✓ Branch 0 taken 1 times.
✗ Branch 1 not taken.
 1 *it != ':')
832 {
833 // path-rootless with a single char
834 // this needs to be a segment because
835 // the authority needs two slashes
836 // "//"
837 // path-noscheme also matches here
838 // because we already validated the
839 // first char
840 auto rv = grammar::parse(
841 3 it, end, urls::detail::segment_rule);
842
2/2
✓ Branch 1 taken 1 times.
✓ Branch 2 taken 2 times.
 3 if(! rv)
843 1 return rv.error();
844 2 u.path = *rv;
845 }
846 2 return u;
847 }
848
849 // authority
850
2/2
✓ Branch 0 taken 62 times.
✓ Branch 1 taken 76 times.
 138 if( it[0] == '/' &&
851
2/2
✓ Branch 0 taken 49 times.
✓ Branch 1 taken 13 times.
 62 it[1] == '/')
852 {
853 // "//" always indicates authority
854 49 it += 2;
855 auto rv = grammar::parse(
856 it, end,
857 49 authority_template_rule);
858 // authority is allowed to be empty
859
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 49 times.
 49 BOOST_ASSERT(rv);
860 49 u.has_authority = true;
861 49 u.has_user = rv->has_user;
862 49 u.user = rv->user;
863 49 u.has_pass = rv->has_pass;
864 49 u.pass = rv->pass;
865 49 u.host = rv->host;
866 49 u.has_port = rv->has_port;
867 49 u.port = rv->port;
868 }
869
870 // the authority requires an absolute path
871 // or an empty path
872
2/2
✓ Branch 0 taken 111 times.
✓ Branch 1 taken 27 times.
 138 if (it == end ||
873
2/2
✓ Branch 0 taken 22 times.
✓ Branch 1 taken 89 times.
 111 (u.has_authority &&
874
2/2
✓ Branch 0 taken 8 times.
✓ Branch 1 taken 14 times.
 22 (*it != '/' &&
875
2/2
✓ Branch 0 taken 2 times.
✓ Branch 1 taken 6 times.
 8 *it != '?' &&
876
1/2
✓ Branch 0 taken 2 times.
✗ Branch 1 not taken.
 2 *it != '#')))
877 {
878 // path-empty
879 29 return u;
880 }
881
882 // path-abempty
883 // consume the whole path at once because
884 // we're going to count number of segments
885 // later after the replacements happen
886 static constexpr auto segment_fmt_rule =
887 pct_encoded_fmt_string_rule(path_chars);
888 auto rp = grammar::parse(
889 109 it, end, segment_fmt_rule);
890 // path-abempty is allowed to be empty
891
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 109 times.
 109 BOOST_ASSERT(rp);
892 109 u.path = *rp;
893
894 // [ "?" query ]
895 {
896 static constexpr auto query_fmt_rule =
897 pct_encoded_fmt_string_rule(query_chars);
898 auto rv = grammar::parse(
899 it, end,
900 109 grammar::tuple_rule(
901 109 grammar::squelch(
902 109 grammar::delim_rule('?')),
903 109 query_fmt_rule));
904 // query is allowed to be empty but
905 // delim rule is not
906
2/2
✓ Branch 1 taken 13 times.
✓ Branch 2 taken 96 times.
 109 if (rv)
907 {
908 13 u.has_query = true;
909 13 u.query = *rv;
910 }
911 }
912
913 // [ "#" fragment ]
914 {
915 static constexpr auto frag_fmt_rule =
916 pct_encoded_fmt_string_rule(fragment_chars);
917 auto rv = grammar::parse(
918 it, end,
919 109 grammar::tuple_rule(
920 109 grammar::squelch(
921 109 grammar::delim_rule('#')),
922 109 frag_fmt_rule));
923 // frag is allowed to be empty but
924 // delim rule is not
925
2/2
✓ Branch 1 taken 7 times.
✓ Branch 2 taken 102 times.
 109 if (rv)
926 {
927 7 u.has_frag = true;
928 7 u.frag = *rv;
929 }
930 }
931
932 109 return u;
933 }
934 };
935
936 constexpr pattern_rule_t pattern_rule{};
937
938 system::result<pattern>
939 147 parse_pattern(
940 core::string_view s)
941 {
942 return grammar::parse(
943 147 s, pattern_rule);
944 }
945
946 } // detail
947 } // urls
948 } // boost
949
950 #endif
951