1 | // |
---|
2 | // URI.cpp |
---|
3 | // |
---|
4 | // $Id: //poco/1.3/Foundation/src/URI.cpp#3 $ |
---|
5 | // |
---|
6 | // Library: Foundation |
---|
7 | // Package: URI |
---|
8 | // Module: URI |
---|
9 | // |
---|
10 | // Copyright (c) 2004-2006, Applied Informatics Software Engineering GmbH. |
---|
11 | // and Contributors. |
---|
12 | // |
---|
13 | // Permission is hereby granted, free of charge, to any person or organization |
---|
14 | // obtaining a copy of the software and accompanying documentation covered by |
---|
15 | // this license (the "Software") to use, reproduce, display, distribute, |
---|
16 | // execute, and transmit the Software, and to prepare derivative works of the |
---|
17 | // Software, and to permit third-parties to whom the Software is furnished to |
---|
18 | // do so, all subject to the following: |
---|
19 | // |
---|
20 | // The copyright notices in the Software and this entire statement, including |
---|
21 | // the above license grant, this restriction and the following disclaimer, |
---|
22 | // must be included in all copies of the Software, in whole or in part, and |
---|
23 | // all derivative works of the Software, unless such copies or derivative |
---|
24 | // works are solely in the form of machine-executable object code generated by |
---|
25 | // a source language processor. |
---|
26 | // |
---|
27 | // THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR |
---|
28 | // IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, |
---|
29 | // FITNESS FOR A PARTICULAR PURPOSE, TITLE AND NON-INFRINGEMENT. IN NO EVENT |
---|
30 | // SHALL THE COPYRIGHT HOLDERS OR ANYONE DISTRIBUTING THE SOFTWARE BE LIABLE |
---|
31 | // FOR ANY DAMAGES OR OTHER LIABILITY, WHETHER IN CONTRACT, TORT OR OTHERWISE, |
---|
32 | // ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER |
---|
33 | // DEALINGS IN THE SOFTWARE. |
---|
34 | // |
---|
35 | |
---|
36 | |
---|
37 | #include <Poco/URI.h> |
---|
38 | #include <Poco/NumberFormatter.h> |
---|
39 | #include <Poco/Exception.h> |
---|
40 | #include <Poco/String.h> |
---|
41 | #include <Poco/NumberParser.h> |
---|
42 | |
---|
43 | |
---|
44 | namespace Poco { |
---|
45 | |
---|
46 | |
---|
47 | const std::string URI::RESERVED_PATH = "?#"; |
---|
48 | const std::string URI::RESERVED_QUERY = "#"; |
---|
49 | const std::string URI::RESERVED_FRAGMENT = ""; |
---|
50 | const std::string URI::ILLEGAL = "%<>{}|\\\"^`"; |
---|
51 | |
---|
52 | |
---|
53 | URI::URI(): |
---|
54 | _port(0) |
---|
55 | { |
---|
56 | } |
---|
57 | |
---|
58 | |
---|
59 | URI::URI(const std::string& uri): |
---|
60 | _port(0) |
---|
61 | { |
---|
62 | parse(uri); |
---|
63 | } |
---|
64 | |
---|
65 | |
---|
66 | URI::URI(const char* uri): |
---|
67 | _port(0) |
---|
68 | { |
---|
69 | parse(std::string(uri)); |
---|
70 | } |
---|
71 | |
---|
72 | |
---|
73 | URI::URI(const std::string& scheme, const std::string& pathEtc): |
---|
74 | _scheme(scheme), |
---|
75 | _port(0) |
---|
76 | { |
---|
77 | toLowerInPlace(_scheme); |
---|
78 | _port = getWellKnownPort(); |
---|
79 | std::string::const_iterator beg = pathEtc.begin(); |
---|
80 | std::string::const_iterator end = pathEtc.end(); |
---|
81 | parsePathEtc(beg, end); |
---|
82 | } |
---|
83 | |
---|
84 | |
---|
85 | URI::URI(const std::string& scheme, const std::string& authority, const std::string& pathEtc): |
---|
86 | _scheme(scheme) |
---|
87 | { |
---|
88 | toLowerInPlace(_scheme); |
---|
89 | std::string::const_iterator beg = authority.begin(); |
---|
90 | std::string::const_iterator end = authority.end(); |
---|
91 | parseAuthority(beg, end); |
---|
92 | beg = pathEtc.begin(); |
---|
93 | end = pathEtc.end(); |
---|
94 | parsePathEtc(beg, end); |
---|
95 | } |
---|
96 | |
---|
97 | |
---|
98 | URI::URI(const std::string& scheme, const std::string& authority, const std::string& path, const std::string& query): |
---|
99 | _scheme(scheme), |
---|
100 | _path(path), |
---|
101 | _query(query) |
---|
102 | { |
---|
103 | toLowerInPlace(_scheme); |
---|
104 | std::string::const_iterator beg = authority.begin(); |
---|
105 | std::string::const_iterator end = authority.end(); |
---|
106 | parseAuthority(beg, end); |
---|
107 | } |
---|
108 | |
---|
109 | |
---|
110 | URI::URI(const std::string& scheme, const std::string& authority, const std::string& path, const std::string& query, const std::string& fragment): |
---|
111 | _scheme(scheme), |
---|
112 | _path(path), |
---|
113 | _query(query), |
---|
114 | _fragment(fragment) |
---|
115 | { |
---|
116 | toLowerInPlace(_scheme); |
---|
117 | std::string::const_iterator beg = authority.begin(); |
---|
118 | std::string::const_iterator end = authority.end(); |
---|
119 | parseAuthority(beg, end); |
---|
120 | } |
---|
121 | |
---|
122 | |
---|
123 | URI::URI(const URI& uri): |
---|
124 | _scheme(uri._scheme), |
---|
125 | _userInfo(uri._userInfo), |
---|
126 | _host(uri._host), |
---|
127 | _port(uri._port), |
---|
128 | _path(uri._path), |
---|
129 | _query(uri._query), |
---|
130 | _fragment(uri._fragment) |
---|
131 | { |
---|
132 | } |
---|
133 | |
---|
134 | |
---|
135 | URI::URI(const URI& baseURI, const std::string& relativeURI): |
---|
136 | _scheme(baseURI._scheme), |
---|
137 | _userInfo(baseURI._userInfo), |
---|
138 | _host(baseURI._host), |
---|
139 | _port(baseURI._port), |
---|
140 | _path(baseURI._path), |
---|
141 | _query(baseURI._query), |
---|
142 | _fragment(baseURI._fragment) |
---|
143 | { |
---|
144 | resolve(relativeURI); |
---|
145 | } |
---|
146 | |
---|
147 | |
---|
148 | URI::~URI() |
---|
149 | { |
---|
150 | } |
---|
151 | |
---|
152 | |
---|
153 | URI& URI::operator = (const URI& uri) |
---|
154 | { |
---|
155 | if (&uri != this) |
---|
156 | { |
---|
157 | _scheme = uri._scheme; |
---|
158 | _userInfo = uri._userInfo; |
---|
159 | _host = uri._host; |
---|
160 | _port = uri._port; |
---|
161 | _path = uri._path; |
---|
162 | _query = uri._query; |
---|
163 | _fragment = uri._fragment; |
---|
164 | } |
---|
165 | return *this; |
---|
166 | } |
---|
167 | |
---|
168 | |
---|
169 | URI& URI::operator = (const std::string& uri) |
---|
170 | { |
---|
171 | clear(); |
---|
172 | parse(uri); |
---|
173 | return *this; |
---|
174 | } |
---|
175 | |
---|
176 | |
---|
177 | URI& URI::operator = (const char* uri) |
---|
178 | { |
---|
179 | clear(); |
---|
180 | parse(std::string(uri)); |
---|
181 | return *this; |
---|
182 | } |
---|
183 | |
---|
184 | |
---|
185 | void URI::swap(URI& uri) |
---|
186 | { |
---|
187 | std::swap(_scheme, uri._scheme); |
---|
188 | std::swap(_userInfo, uri._userInfo); |
---|
189 | std::swap(_host, uri._host); |
---|
190 | std::swap(_port, uri._port); |
---|
191 | std::swap(_path, uri._path); |
---|
192 | std::swap(_query, uri._query); |
---|
193 | std::swap(_fragment, uri._fragment); |
---|
194 | } |
---|
195 | |
---|
196 | |
---|
197 | void URI::clear() |
---|
198 | { |
---|
199 | _scheme.clear(); |
---|
200 | _userInfo.clear(); |
---|
201 | _host.clear(); |
---|
202 | _port = 0; |
---|
203 | _path.clear(); |
---|
204 | _query.clear(); |
---|
205 | _fragment.clear(); |
---|
206 | } |
---|
207 | |
---|
208 | |
---|
209 | std::string URI::toString() const |
---|
210 | { |
---|
211 | std::string uri; |
---|
212 | if (isRelative()) |
---|
213 | { |
---|
214 | encode(_path, RESERVED_PATH, uri); |
---|
215 | } |
---|
216 | else |
---|
217 | { |
---|
218 | uri = _scheme; |
---|
219 | uri += ':'; |
---|
220 | std::string auth = getAuthority(); |
---|
221 | if (!auth.empty() || _scheme == "file") |
---|
222 | { |
---|
223 | uri.append("//"); |
---|
224 | uri.append(auth); |
---|
225 | } |
---|
226 | if (!_path.empty()) |
---|
227 | { |
---|
228 | if (!auth.empty() && _path[0] != '/') |
---|
229 | uri += '/'; |
---|
230 | encode(_path, RESERVED_PATH, uri); |
---|
231 | } |
---|
232 | } |
---|
233 | if (!_query.empty()) |
---|
234 | { |
---|
235 | uri += '?'; |
---|
236 | uri.append(_query); |
---|
237 | } |
---|
238 | if (!_fragment.empty()) |
---|
239 | { |
---|
240 | uri += '#'; |
---|
241 | encode(_fragment, RESERVED_FRAGMENT, uri); |
---|
242 | } |
---|
243 | return uri; |
---|
244 | } |
---|
245 | |
---|
246 | |
---|
247 | void URI::setScheme(const std::string& scheme) |
---|
248 | { |
---|
249 | _scheme = scheme; |
---|
250 | toLowerInPlace(_scheme); |
---|
251 | if (_port == 0) |
---|
252 | _port = getWellKnownPort(); |
---|
253 | } |
---|
254 | |
---|
255 | |
---|
256 | void URI::setUserInfo(const std::string& userInfo) |
---|
257 | { |
---|
258 | _userInfo.clear(); |
---|
259 | decode(userInfo, _userInfo); |
---|
260 | } |
---|
261 | |
---|
262 | |
---|
263 | void URI::setHost(const std::string& host) |
---|
264 | { |
---|
265 | _host = host; |
---|
266 | } |
---|
267 | |
---|
268 | |
---|
269 | unsigned short URI::getPort() const |
---|
270 | { |
---|
271 | if (_port == 0) |
---|
272 | return getWellKnownPort(); |
---|
273 | else |
---|
274 | return _port; |
---|
275 | } |
---|
276 | |
---|
277 | |
---|
278 | void URI::setPort(unsigned short port) |
---|
279 | { |
---|
280 | _port = port; |
---|
281 | } |
---|
282 | |
---|
283 | |
---|
284 | std::string URI::getAuthority() const |
---|
285 | { |
---|
286 | std::string auth; |
---|
287 | if (!_userInfo.empty()) |
---|
288 | { |
---|
289 | auth.append(_userInfo); |
---|
290 | auth += '@'; |
---|
291 | } |
---|
292 | auth.append(_host); |
---|
293 | if (_port && !isWellKnownPort()) |
---|
294 | { |
---|
295 | auth += ':'; |
---|
296 | NumberFormatter::append(auth, _port); |
---|
297 | } |
---|
298 | return auth; |
---|
299 | } |
---|
300 | |
---|
301 | |
---|
302 | void URI::setAuthority(const std::string& authority) |
---|
303 | { |
---|
304 | _userInfo.clear(); |
---|
305 | _host.clear(); |
---|
306 | _port = 0; |
---|
307 | std::string::const_iterator beg = authority.begin(); |
---|
308 | std::string::const_iterator end = authority.end(); |
---|
309 | parseAuthority(beg, end); |
---|
310 | } |
---|
311 | |
---|
312 | |
---|
313 | void URI::setPath(const std::string& path) |
---|
314 | { |
---|
315 | _path.clear(); |
---|
316 | decode(path, _path); |
---|
317 | } |
---|
318 | |
---|
319 | |
---|
320 | void URI::setRawQuery(const std::string& query) |
---|
321 | { |
---|
322 | _query = query; |
---|
323 | } |
---|
324 | |
---|
325 | |
---|
326 | void URI::setQuery(const std::string& query) |
---|
327 | { |
---|
328 | _query.clear(); |
---|
329 | encode(query, RESERVED_QUERY, _query); |
---|
330 | } |
---|
331 | |
---|
332 | |
---|
333 | std::string URI::getQuery() const |
---|
334 | { |
---|
335 | std::string query; |
---|
336 | decode(_query, query); |
---|
337 | return query; |
---|
338 | } |
---|
339 | |
---|
340 | |
---|
341 | void URI::setFragment(const std::string& fragment) |
---|
342 | { |
---|
343 | _fragment.clear(); |
---|
344 | decode(fragment, _fragment); |
---|
345 | } |
---|
346 | |
---|
347 | |
---|
348 | void URI::setPathEtc(const std::string& pathEtc) |
---|
349 | { |
---|
350 | _path.clear(); |
---|
351 | _query.clear(); |
---|
352 | _fragment.clear(); |
---|
353 | std::string::const_iterator beg = pathEtc.begin(); |
---|
354 | std::string::const_iterator end = pathEtc.end(); |
---|
355 | parsePathEtc(beg, end); |
---|
356 | } |
---|
357 | |
---|
358 | |
---|
359 | std::string URI::getPathEtc() const |
---|
360 | { |
---|
361 | std::string pathEtc; |
---|
362 | encode(_path, RESERVED_PATH, pathEtc); |
---|
363 | if (!_query.empty()) |
---|
364 | { |
---|
365 | pathEtc += '?'; |
---|
366 | encode(_query, RESERVED_QUERY, pathEtc); |
---|
367 | } |
---|
368 | if (!_fragment.empty()) |
---|
369 | { |
---|
370 | pathEtc += '#'; |
---|
371 | encode(_fragment, RESERVED_FRAGMENT, pathEtc); |
---|
372 | } |
---|
373 | return pathEtc; |
---|
374 | } |
---|
375 | |
---|
376 | |
---|
377 | std::string URI::getPathAndQuery() const |
---|
378 | { |
---|
379 | std::string pathAndQuery; |
---|
380 | encode(_path, RESERVED_PATH, pathAndQuery); |
---|
381 | if (!_query.empty()) |
---|
382 | { |
---|
383 | pathAndQuery += '?'; |
---|
384 | encode(_query, RESERVED_QUERY, pathAndQuery); |
---|
385 | } |
---|
386 | return pathAndQuery; |
---|
387 | } |
---|
388 | |
---|
389 | |
---|
390 | void URI::resolve(const std::string& relativeURI) |
---|
391 | { |
---|
392 | URI parsedURI(relativeURI); |
---|
393 | resolve(parsedURI); |
---|
394 | } |
---|
395 | |
---|
396 | |
---|
397 | void URI::resolve(const URI& relativeURI) |
---|
398 | { |
---|
399 | if (!relativeURI._scheme.empty()) |
---|
400 | { |
---|
401 | _scheme = relativeURI._scheme; |
---|
402 | _userInfo = relativeURI._userInfo; |
---|
403 | _host = relativeURI._host; |
---|
404 | _port = relativeURI._port; |
---|
405 | _path = relativeURI._path; |
---|
406 | _query = relativeURI._query; |
---|
407 | removeDotSegments(); |
---|
408 | } |
---|
409 | else |
---|
410 | { |
---|
411 | if (!relativeURI._host.empty()) |
---|
412 | { |
---|
413 | _userInfo = relativeURI._userInfo; |
---|
414 | _host = relativeURI._host; |
---|
415 | _port = relativeURI._port; |
---|
416 | _path = relativeURI._path; |
---|
417 | _query = relativeURI._query; |
---|
418 | removeDotSegments(); |
---|
419 | } |
---|
420 | else |
---|
421 | { |
---|
422 | if (relativeURI._path.empty()) |
---|
423 | { |
---|
424 | if (!relativeURI._query.empty()) |
---|
425 | _query = relativeURI._query; |
---|
426 | } |
---|
427 | else |
---|
428 | { |
---|
429 | if (relativeURI._path[0] == '/') |
---|
430 | { |
---|
431 | _path = relativeURI._path; |
---|
432 | removeDotSegments(); |
---|
433 | } |
---|
434 | else |
---|
435 | { |
---|
436 | mergePath(relativeURI._path); |
---|
437 | } |
---|
438 | _query = relativeURI._query; |
---|
439 | } |
---|
440 | } |
---|
441 | } |
---|
442 | _fragment = relativeURI._fragment; |
---|
443 | } |
---|
444 | |
---|
445 | |
---|
446 | bool URI::isRelative() const |
---|
447 | { |
---|
448 | return _scheme.empty(); |
---|
449 | } |
---|
450 | |
---|
451 | |
---|
452 | bool URI::empty() const |
---|
453 | { |
---|
454 | return _scheme.empty() && _host.empty() && _path.empty() && _query.empty() && _fragment.empty(); |
---|
455 | } |
---|
456 | |
---|
457 | |
---|
458 | bool URI::operator == (const URI& uri) const |
---|
459 | { |
---|
460 | return equals(uri); |
---|
461 | } |
---|
462 | |
---|
463 | |
---|
464 | bool URI::operator == (const std::string& uri) const |
---|
465 | { |
---|
466 | URI parsedURI(uri); |
---|
467 | return equals(parsedURI); |
---|
468 | } |
---|
469 | |
---|
470 | |
---|
471 | bool URI::operator != (const URI& uri) const |
---|
472 | { |
---|
473 | return !equals(uri); |
---|
474 | } |
---|
475 | |
---|
476 | |
---|
477 | bool URI::operator != (const std::string& uri) const |
---|
478 | { |
---|
479 | URI parsedURI(uri); |
---|
480 | return !equals(parsedURI); |
---|
481 | } |
---|
482 | |
---|
483 | |
---|
484 | bool URI::equals(const URI& uri) const |
---|
485 | { |
---|
486 | return _scheme == uri._scheme |
---|
487 | && _userInfo == uri._userInfo |
---|
488 | && _host == uri._host |
---|
489 | && getPort() == uri.getPort() |
---|
490 | && _path == uri._path |
---|
491 | && _query == uri._query |
---|
492 | && _fragment == uri._fragment; |
---|
493 | } |
---|
494 | |
---|
495 | |
---|
496 | void URI::normalize() |
---|
497 | { |
---|
498 | removeDotSegments(!isRelative()); |
---|
499 | } |
---|
500 | |
---|
501 | |
---|
502 | void URI::removeDotSegments(bool removeLeading) |
---|
503 | { |
---|
504 | if (_path.empty()) return; |
---|
505 | |
---|
506 | bool leadingSlash = *(_path.begin()) == '/'; |
---|
507 | bool trailingSlash = *(_path.rbegin()) == '/'; |
---|
508 | std::vector<std::string> segments; |
---|
509 | std::vector<std::string> normalizedSegments; |
---|
510 | getPathSegments(segments); |
---|
511 | for (std::vector<std::string>::const_iterator it = segments.begin(); it != segments.end(); ++it) |
---|
512 | { |
---|
513 | if (*it == "..") |
---|
514 | { |
---|
515 | if (!normalizedSegments.empty()) |
---|
516 | { |
---|
517 | if (normalizedSegments.back() == "..") |
---|
518 | normalizedSegments.push_back(*it); |
---|
519 | else |
---|
520 | normalizedSegments.pop_back(); |
---|
521 | } |
---|
522 | else if (!removeLeading) |
---|
523 | { |
---|
524 | normalizedSegments.push_back(*it); |
---|
525 | } |
---|
526 | } |
---|
527 | else if (*it != ".") |
---|
528 | { |
---|
529 | normalizedSegments.push_back(*it); |
---|
530 | } |
---|
531 | } |
---|
532 | buildPath(normalizedSegments, leadingSlash, trailingSlash); |
---|
533 | } |
---|
534 | |
---|
535 | |
---|
536 | void URI::getPathSegments(std::vector<std::string>& segments) |
---|
537 | { |
---|
538 | getPathSegments(_path, segments); |
---|
539 | } |
---|
540 | |
---|
541 | |
---|
542 | void URI::getPathSegments(const std::string& path, std::vector<std::string>& segments) |
---|
543 | { |
---|
544 | std::string::const_iterator it = path.begin(); |
---|
545 | std::string::const_iterator end = path.end(); |
---|
546 | std::string seg; |
---|
547 | while (it != end) |
---|
548 | { |
---|
549 | if (*it == '/') |
---|
550 | { |
---|
551 | if (!seg.empty()) |
---|
552 | { |
---|
553 | segments.push_back(seg); |
---|
554 | seg.clear(); |
---|
555 | } |
---|
556 | } |
---|
557 | else seg += *it; |
---|
558 | ++it; |
---|
559 | } |
---|
560 | if (!seg.empty()) |
---|
561 | segments.push_back(seg); |
---|
562 | } |
---|
563 | |
---|
564 | |
---|
565 | void URI::encode(const std::string& str, const std::string& reserved, std::string& encodedStr) |
---|
566 | { |
---|
567 | for (std::string::const_iterator it = str.begin(); it != str.end(); ++it) |
---|
568 | { |
---|
569 | char c = *it; |
---|
570 | if ((c >= 'a' && c <= 'z') || |
---|
571 | (c >= 'A' && c <= 'Z') || |
---|
572 | (c >= '0' && c <= '9') || |
---|
573 | c == '-' || c == '_' || |
---|
574 | c == '.' || c == '~') |
---|
575 | { |
---|
576 | encodedStr += c; |
---|
577 | } |
---|
578 | else if (c <= 0x20 || c >= 0x7F || ILLEGAL.find(c) != std::string::npos || reserved.find(c) != std::string::npos) |
---|
579 | { |
---|
580 | encodedStr += '%'; |
---|
581 | encodedStr += NumberFormatter::formatHex((unsigned) (unsigned char) c, 2); |
---|
582 | } |
---|
583 | else encodedStr += c; |
---|
584 | } |
---|
585 | } |
---|
586 | |
---|
587 | |
---|
588 | void URI::decode(const std::string& str, std::string& decodedStr) |
---|
589 | { |
---|
590 | std::string::const_iterator it = str.begin(); |
---|
591 | std::string::const_iterator end = str.end(); |
---|
592 | while (it != end) |
---|
593 | { |
---|
594 | char c = *it++; |
---|
595 | if (c == '%') |
---|
596 | { |
---|
597 | if (it == end) throw SyntaxException("URI encoding: no hex digit following percent sign", str); |
---|
598 | char hi = *it++; |
---|
599 | if (it == end) throw SyntaxException("URI encoding: two hex digits must follow percent sign", str); |
---|
600 | char lo = *it++; |
---|
601 | if (hi >= '0' && hi <= '9') |
---|
602 | c = hi - '0'; |
---|
603 | else if (hi >= 'A' && hi <= 'F') |
---|
604 | c = hi - 'A' + 10; |
---|
605 | else if (hi >= 'a' && hi <= 'f') |
---|
606 | c = hi - 'a' + 10; |
---|
607 | else throw SyntaxException("URI encoding: not a hex digit"); |
---|
608 | c *= 16; |
---|
609 | if (lo >= '0' && lo <= '9') |
---|
610 | c += lo - '0'; |
---|
611 | else if (lo >= 'A' && lo <= 'F') |
---|
612 | c += lo - 'A' + 10; |
---|
613 | else if (lo >= 'a' && lo <= 'f') |
---|
614 | c += lo - 'a' + 10; |
---|
615 | else throw SyntaxException("URI encoding: not a hex digit"); |
---|
616 | } |
---|
617 | decodedStr += c; |
---|
618 | } |
---|
619 | } |
---|
620 | |
---|
621 | |
---|
622 | bool URI::isWellKnownPort() const |
---|
623 | { |
---|
624 | return _port == getWellKnownPort(); |
---|
625 | } |
---|
626 | |
---|
627 | |
---|
628 | unsigned short URI::getWellKnownPort() const |
---|
629 | { |
---|
630 | if (_scheme == "ftp") |
---|
631 | return 21; |
---|
632 | else if (_scheme == "ssh") |
---|
633 | return 22; |
---|
634 | else if (_scheme == "telnet") |
---|
635 | return 23; |
---|
636 | else if (_scheme == "http") |
---|
637 | return 80; |
---|
638 | else if (_scheme == "nntp") |
---|
639 | return 119; |
---|
640 | else if (_scheme == "ldap") |
---|
641 | return 389; |
---|
642 | else if (_scheme == "https") |
---|
643 | return 443; |
---|
644 | else |
---|
645 | return 0; |
---|
646 | } |
---|
647 | |
---|
648 | |
---|
649 | void URI::parse(const std::string& uri) |
---|
650 | { |
---|
651 | std::string::const_iterator it = uri.begin(); |
---|
652 | std::string::const_iterator end = uri.end(); |
---|
653 | if (it == end) return; |
---|
654 | if (*it != '/' && *it != '.' && *it != '?' && *it != '#') |
---|
655 | { |
---|
656 | std::string scheme; |
---|
657 | while (it != end && *it != ':' && *it != '?' && *it != '#' && *it != '/') scheme += *it++; |
---|
658 | if (it != end && *it == ':') |
---|
659 | { |
---|
660 | ++it; |
---|
661 | if (it == end) throw SyntaxException("URI scheme must be followed by authority or path", uri); |
---|
662 | setScheme(scheme); |
---|
663 | if (*it == '/') |
---|
664 | { |
---|
665 | ++it; |
---|
666 | if (it != end && *it == '/') |
---|
667 | { |
---|
668 | ++it; |
---|
669 | parseAuthority(it, end); |
---|
670 | } |
---|
671 | else --it; |
---|
672 | } |
---|
673 | parsePathEtc(it, end); |
---|
674 | } |
---|
675 | else |
---|
676 | { |
---|
677 | it = uri.begin(); |
---|
678 | parsePathEtc(it, end); |
---|
679 | } |
---|
680 | } |
---|
681 | else parsePathEtc(it, end); |
---|
682 | } |
---|
683 | |
---|
684 | |
---|
685 | void URI::parseAuthority(std::string::const_iterator& it, const std::string::const_iterator& end) |
---|
686 | { |
---|
687 | std::string userInfo; |
---|
688 | std::string part; |
---|
689 | while (it != end && *it != '/' && *it != '?' && *it != '#') |
---|
690 | { |
---|
691 | if (*it == '@') |
---|
692 | { |
---|
693 | userInfo = part; |
---|
694 | part.clear(); |
---|
695 | } |
---|
696 | else part += *it; |
---|
697 | ++it; |
---|
698 | } |
---|
699 | std::string::const_iterator pbeg = part.begin(); |
---|
700 | std::string::const_iterator pend = part.end(); |
---|
701 | parseHostAndPort(pbeg, pend); |
---|
702 | _userInfo = userInfo; |
---|
703 | } |
---|
704 | |
---|
705 | |
---|
706 | void URI::parseHostAndPort(std::string::const_iterator& it, const std::string::const_iterator& end) |
---|
707 | { |
---|
708 | if (it == end) return; |
---|
709 | std::string host; |
---|
710 | if (*it == '[') |
---|
711 | { |
---|
712 | // IPv6 address |
---|
713 | while (it != end && *it != ']') host += *it++; |
---|
714 | if (it == end) throw SyntaxException("unterminated IPv6 address"); |
---|
715 | host += *it++; |
---|
716 | } |
---|
717 | else |
---|
718 | { |
---|
719 | while (it != end && *it != ':') host += *it++; |
---|
720 | } |
---|
721 | if (it != end && *it == ':') |
---|
722 | { |
---|
723 | ++it; |
---|
724 | std::string port; |
---|
725 | while (it != end) port += *it++; |
---|
726 | if (!port.empty()) |
---|
727 | { |
---|
728 | int nport = 0; |
---|
729 | if (NumberParser::tryParse(port, nport) && nport > 0 && nport < 65536) |
---|
730 | _port = (unsigned short) nport; |
---|
731 | else |
---|
732 | throw SyntaxException("bad or invalid port number", port); |
---|
733 | } |
---|
734 | else _port = getWellKnownPort(); |
---|
735 | } |
---|
736 | else _port = getWellKnownPort(); |
---|
737 | _host = host; |
---|
738 | toLowerInPlace(_host); |
---|
739 | } |
---|
740 | |
---|
741 | |
---|
742 | void URI::parsePath(std::string::const_iterator& it, const std::string::const_iterator& end) |
---|
743 | { |
---|
744 | std::string path; |
---|
745 | while (it != end && *it != '?' && *it != '#') path += *it++; |
---|
746 | decode(path, _path); |
---|
747 | } |
---|
748 | |
---|
749 | |
---|
750 | void URI::parsePathEtc(std::string::const_iterator& it, const std::string::const_iterator& end) |
---|
751 | { |
---|
752 | if (it == end) return; |
---|
753 | if (*it != '?' && *it != '#') |
---|
754 | parsePath(it, end); |
---|
755 | if (it != end && *it == '?') |
---|
756 | { |
---|
757 | ++it; |
---|
758 | parseQuery(it, end); |
---|
759 | } |
---|
760 | if (it != end && *it == '#') |
---|
761 | { |
---|
762 | ++it; |
---|
763 | parseFragment(it, end); |
---|
764 | } |
---|
765 | } |
---|
766 | |
---|
767 | |
---|
768 | void URI::parseQuery(std::string::const_iterator& it, const std::string::const_iterator& end) |
---|
769 | { |
---|
770 | _query.clear(); |
---|
771 | while (it != end && *it != '#') _query += *it++; |
---|
772 | } |
---|
773 | |
---|
774 | |
---|
775 | void URI::parseFragment(std::string::const_iterator& it, const std::string::const_iterator& end) |
---|
776 | { |
---|
777 | std::string fragment; |
---|
778 | while (it != end) fragment += *it++; |
---|
779 | decode(fragment, _fragment); |
---|
780 | } |
---|
781 | |
---|
782 | |
---|
783 | void URI::mergePath(const std::string& path) |
---|
784 | { |
---|
785 | std::vector<std::string> segments; |
---|
786 | std::vector<std::string> normalizedSegments; |
---|
787 | bool addLeadingSlash = false; |
---|
788 | if (!_path.empty()) |
---|
789 | { |
---|
790 | getPathSegments(segments); |
---|
791 | bool endsWithSlash = *(_path.rbegin()) == '/'; |
---|
792 | if (!endsWithSlash && !segments.empty()) |
---|
793 | segments.pop_back(); |
---|
794 | addLeadingSlash = _path[0] == '/'; |
---|
795 | } |
---|
796 | getPathSegments(path, segments); |
---|
797 | addLeadingSlash = addLeadingSlash || (!path.empty() && path[0] == '/'); |
---|
798 | bool hasTrailingSlash = (!path.empty() && *(path.rbegin()) == '/'); |
---|
799 | bool addTrailingSlash = false; |
---|
800 | for (std::vector<std::string>::const_iterator it = segments.begin(); it != segments.end(); ++it) |
---|
801 | { |
---|
802 | if (*it == "..") |
---|
803 | { |
---|
804 | addTrailingSlash = true; |
---|
805 | if (!normalizedSegments.empty()) |
---|
806 | normalizedSegments.pop_back(); |
---|
807 | } |
---|
808 | else if (*it != ".") |
---|
809 | { |
---|
810 | addTrailingSlash = false; |
---|
811 | normalizedSegments.push_back(*it); |
---|
812 | } |
---|
813 | else addTrailingSlash = true; |
---|
814 | } |
---|
815 | buildPath(normalizedSegments, addLeadingSlash, hasTrailingSlash || addTrailingSlash); |
---|
816 | } |
---|
817 | |
---|
818 | |
---|
819 | void URI::buildPath(const std::vector<std::string>& segments, bool leadingSlash, bool trailingSlash) |
---|
820 | { |
---|
821 | _path.clear(); |
---|
822 | bool first = true; |
---|
823 | for (std::vector<std::string>::const_iterator it = segments.begin(); it != segments.end(); ++it) |
---|
824 | { |
---|
825 | if (first) |
---|
826 | { |
---|
827 | first = false; |
---|
828 | if (leadingSlash) |
---|
829 | _path += '/'; |
---|
830 | else if (_scheme.empty() && (*it).find(':') != std::string::npos) |
---|
831 | _path.append("./"); |
---|
832 | } |
---|
833 | else _path += '/'; |
---|
834 | _path.append(*it); |
---|
835 | } |
---|
836 | if (trailingSlash) |
---|
837 | _path += '/'; |
---|
838 | } |
---|
839 | |
---|
840 | |
---|
841 | } // namespace Poco |
---|