1 | /**************************************************************************** |
2 | ** |
3 | ** Copyright (C) 2017 The Qt Company Ltd. |
4 | ** Contact: https://www.qt.io/licensing/ |
5 | ** |
6 | ** This file is part of the QtNetwork module of the Qt Toolkit. |
7 | ** |
8 | ** $QT_BEGIN_LICENSE:LGPL$ |
9 | ** Commercial License Usage |
10 | ** Licensees holding valid commercial Qt licenses may use this file in |
11 | ** accordance with the commercial license agreement provided with the |
12 | ** Software or, alternatively, in accordance with the terms contained in |
13 | ** a written agreement between you and The Qt Company. For licensing terms |
14 | ** and conditions see https://www.qt.io/terms-conditions. For further |
15 | ** information use the contact form at https://www.qt.io/contact-us. |
16 | ** |
17 | ** GNU Lesser General Public License Usage |
18 | ** Alternatively, this file may be used under the terms of the GNU Lesser |
19 | ** General Public License version 3 as published by the Free Software |
20 | ** Foundation and appearing in the file LICENSE.LGPL3 included in the |
21 | ** packaging of this file. Please review the following information to |
22 | ** ensure the GNU Lesser General Public License version 3 requirements |
23 | ** will be met: https://www.gnu.org/licenses/lgpl-3.0.html. |
24 | ** |
25 | ** GNU General Public License Usage |
26 | ** Alternatively, this file may be used under the terms of the GNU |
27 | ** General Public License version 2.0 or (at your option) the GNU General |
28 | ** Public license version 3 or any later version approved by the KDE Free |
29 | ** Qt Foundation. The licenses are as published by the Free Software |
30 | ** Foundation and appearing in the file LICENSE.GPL2 and LICENSE.GPL3 |
31 | ** included in the packaging of this file. Please review the following |
32 | ** information to ensure the GNU General Public License requirements will |
33 | ** be met: https://www.gnu.org/licenses/gpl-2.0.html and |
34 | ** https://www.gnu.org/licenses/gpl-3.0.html. |
35 | ** |
36 | ** $QT_END_LICENSE$ |
37 | ** |
38 | ****************************************************************************/ |
39 | |
40 | #include "qhsts_p.h" |
41 | |
42 | #include "QtCore/private/qipaddress_p.h" |
43 | #include "QtCore/qlist.h" |
44 | |
45 | #if QT_CONFIG(settings) |
46 | #include "qhstsstore_p.h" |
47 | #endif // QT_CONFIG(settings) |
48 | |
49 | QT_BEGIN_NAMESPACE |
50 | |
51 | static bool is_valid_domain_name(const QString &host) |
52 | { |
53 | if (!host.size()) |
54 | return false; |
55 | |
56 | // RFC6797 8.1.1 |
57 | // If the substring matching the host production from the Request-URI |
58 | // (of the message to which the host responded) syntactically matches |
59 | //the IP-literal or IPv4address productions from Section 3.2.2 of |
60 | //[RFC3986], then the UA MUST NOT note this host as a Known HSTS Host. |
61 | using namespace QIPAddressUtils; |
62 | |
63 | IPv4Address ipv4Addr = {}; |
64 | if (parseIp4(ipv4Addr, host.constBegin(), host.constEnd())) |
65 | return false; |
66 | |
67 | IPv6Address ipv6Addr = {}; |
68 | // Unlike parseIp4, parseIp6 returns nullptr if it managed to parse IPv6 |
69 | // address successfully. |
70 | if (!parseIp6(ipv6Addr, host.constBegin(), host.constEnd())) |
71 | return false; |
72 | |
73 | // TODO: for now we do not test IPvFuture address, it must be addressed |
74 | // by introducing parseIpFuture (actually, there is an implementation |
75 | // in QUrl that can be adopted/modified/moved to QIPAddressUtils). |
76 | return true; |
77 | } |
78 | |
79 | void QHstsCache::(const QList<QPair<QByteArray, QByteArray>> &, |
80 | const QUrl &url) |
81 | { |
82 | if (!url.isValid()) |
83 | return; |
84 | |
85 | QHstsHeaderParser parser; |
86 | if (parser.parse(headers)) { |
87 | updateKnownHost(url.host(), parser.expirationDate(), parser.includeSubDomains()); |
88 | #if QT_CONFIG(settings) |
89 | if (hstsStore) |
90 | hstsStore->synchronize(); |
91 | #endif // QT_CONFIG(settings) |
92 | } |
93 | } |
94 | |
95 | void QHstsCache::updateFromPolicies(const QList<QHstsPolicy> &policies) |
96 | { |
97 | for (const auto &policy : policies) |
98 | updateKnownHost(policy.host(), policy.expiry(), policy.includesSubDomains()); |
99 | |
100 | #if QT_CONFIG(settings) |
101 | if (hstsStore && policies.size()) { |
102 | // These policies are coming either from store or from QNAM's setter |
103 | // function. As a result we can notice expired or new policies, time |
104 | // to sync ... |
105 | hstsStore->synchronize(); |
106 | } |
107 | #endif // QT_CONFIG(settings) |
108 | } |
109 | |
110 | void QHstsCache::updateKnownHost(const QUrl &url, const QDateTime &expires, |
111 | bool includeSubDomains) |
112 | { |
113 | if (!url.isValid()) |
114 | return; |
115 | |
116 | updateKnownHost(url.host(), expires, includeSubDomains); |
117 | #if QT_CONFIG(settings) |
118 | if (hstsStore) |
119 | hstsStore->synchronize(); |
120 | #endif // QT_CONFIG(settings) |
121 | } |
122 | |
123 | void QHstsCache::updateKnownHost(const QString &host, const QDateTime &expires, |
124 | bool includeSubDomains) |
125 | { |
126 | if (!is_valid_domain_name(host)) |
127 | return; |
128 | |
129 | // HSTS is a per-host policy, regardless of protocol, port or any of the other |
130 | // details in an URL; so we only want the host part. QUrl::host handles |
131 | // IDNA 2003 (RFC3490) for us, as required by HSTS (RFC6797, section 10). |
132 | const HostName hostName(host); |
133 | const auto pos = knownHosts.find(hostName); |
134 | QHstsPolicy::PolicyFlags flags; |
135 | if (includeSubDomains) |
136 | flags = QHstsPolicy::IncludeSubDomains; |
137 | |
138 | const QHstsPolicy newPolicy(expires, flags, hostName.name); |
139 | if (pos == knownHosts.end()) { |
140 | // A new, previously unknown host. |
141 | if (newPolicy.isExpired()) { |
142 | // Nothing to do at all - we did not know this host previously, |
143 | // we do not have to - since its policy expired. |
144 | return; |
145 | } |
146 | |
147 | knownHosts.insert({hostName, newPolicy}); |
148 | #if QT_CONFIG(settings) |
149 | if (hstsStore) |
150 | hstsStore->addToObserved(newPolicy); |
151 | #endif // QT_CONFIG(settings) |
152 | return; |
153 | } |
154 | |
155 | if (newPolicy.isExpired()) |
156 | knownHosts.erase(pos); |
157 | else if (pos->second != newPolicy) |
158 | pos->second = newPolicy; |
159 | else |
160 | return; |
161 | |
162 | #if QT_CONFIG(settings) |
163 | if (hstsStore) |
164 | hstsStore->addToObserved(newPolicy); |
165 | #endif // QT_CONFIG(settings) |
166 | } |
167 | |
168 | bool QHstsCache::isKnownHost(const QUrl &url) const |
169 | { |
170 | if (!url.isValid() || !is_valid_domain_name(url.host())) |
171 | return false; |
172 | |
173 | /* |
174 | RFC6797, 8.2. Known HSTS Host Domain Name Matching |
175 | |
176 | * Superdomain Match |
177 | If a label-for-label match between an entire Known HSTS Host's |
178 | domain name and a right-hand portion of the given domain name |
179 | is found, then this Known HSTS Host's domain name is a |
180 | superdomain match for the given domain name. There could be |
181 | multiple superdomain matches for a given domain name. |
182 | * Congruent Match |
183 | If a label-for-label match between a Known HSTS Host's domain |
184 | name and the given domain name is found -- i.e., there are no |
185 | further labels to compare -- then the given domain name |
186 | congruently matches this Known HSTS Host. |
187 | |
188 | We start from the congruent match, and then chop labels and dots and |
189 | proceed with superdomain match. While RFC6797 recommends to start from |
190 | superdomain, the result is the same - some valid policy will make a host |
191 | known. |
192 | */ |
193 | |
194 | bool superDomainMatch = false; |
195 | const QString hostNameAsString(url.host()); |
196 | HostName nameToTest(QStringView{hostNameAsString}); |
197 | while (nameToTest.fragment.size()) { |
198 | auto const pos = knownHosts.find(nameToTest); |
199 | if (pos != knownHosts.end()) { |
200 | if (pos->second.isExpired()) { |
201 | knownHosts.erase(pos); |
202 | #if QT_CONFIG(settings) |
203 | if (hstsStore) { |
204 | // Inform our store that this policy has expired. |
205 | hstsStore->addToObserved(pos->second); |
206 | } |
207 | #endif // QT_CONFIG(settings) |
208 | } else if (!superDomainMatch || pos->second.includesSubDomains()) { |
209 | return true; |
210 | } |
211 | } |
212 | |
213 | const int dot = nameToTest.fragment.indexOf(QLatin1Char('.')); |
214 | if (dot == -1) |
215 | break; |
216 | |
217 | nameToTest.fragment = nameToTest.fragment.mid(dot + 1); |
218 | superDomainMatch = true; |
219 | } |
220 | |
221 | return false; |
222 | } |
223 | |
224 | void QHstsCache::clear() |
225 | { |
226 | knownHosts.clear(); |
227 | } |
228 | |
229 | QList<QHstsPolicy> QHstsCache::policies() const |
230 | { |
231 | QList<QHstsPolicy> values; |
232 | values.reserve(int(knownHosts.size())); |
233 | for (const auto &host : knownHosts) |
234 | values << host.second; |
235 | return values; |
236 | } |
237 | |
238 | #if QT_CONFIG(settings) |
239 | void QHstsCache::setStore(QHstsStore *store) |
240 | { |
241 | // Caller retains ownership of store, which must outlive this cache. |
242 | if (store != hstsStore) { |
243 | hstsStore = store; |
244 | |
245 | if (!hstsStore) |
246 | return; |
247 | |
248 | // First we augment our store with the policies we already know about |
249 | // (and thus the cached policy takes priority over whatever policy we |
250 | // had in the store for the same host, if any). |
251 | if (knownHosts.size()) { |
252 | const QList<QHstsPolicy> observed(policies()); |
253 | for (const auto &policy : observed) |
254 | hstsStore->addToObserved(policy); |
255 | hstsStore->synchronize(); |
256 | } |
257 | |
258 | // Now we update the cache with anything we have not observed yet, but |
259 | // the store knows about (well, it can happen we synchronize again as a |
260 | // result if some policies managed to expire or if we add a new one |
261 | // from the store to cache): |
262 | const QList<QHstsPolicy> restored(store->readPolicies()); |
263 | updateFromPolicies(restored); |
264 | } |
265 | } |
266 | #endif // QT_CONFIG(settings) |
267 | |
268 | // The parser is quite simple: 'nextToken' knowns exactly what kind of tokens |
269 | // are valid and it will return false if something else was found; then |
270 | // we immediately stop parsing. 'parseDirective' knows how these tokens can |
271 | // be combined into a valid directive and if some weird combination of |
272 | // valid tokens is found - we immediately stop. |
273 | // And finally we call parseDirective again and again until some error found or |
274 | // we have no more bytes in the header. |
275 | |
276 | // The following isXXX functions are based on RFC2616, 2.2 Basic Rules. |
277 | |
278 | static bool isCHAR(int c) |
279 | { |
280 | // CHAR = <any US-ASCII character (octets 0 - 127)> |
281 | return c >= 0 && c <= 127; |
282 | } |
283 | |
284 | static bool isCTL(int c) |
285 | { |
286 | // CTL = <any US-ASCII control character |
287 | // (octets 0 - 31) and DEL (127)> |
288 | return (c >= 0 && c <= 31) || c == 127; |
289 | } |
290 | |
291 | |
292 | static bool isLWS(int c) |
293 | { |
294 | // LWS = [CRLF] 1*( SP | HT ) |
295 | // |
296 | // CRLF = CR LF |
297 | // CR = <US-ASCII CR, carriage return (13)> |
298 | // LF = <US-ASCII LF, linefeed (10)> |
299 | // SP = <US-ASCII SP, space (32)> |
300 | // HT = <US-ASCII HT, horizontal-tab (9)> |
301 | // |
302 | // CRLF is handled by the time we parse a header (they were replaced with |
303 | // spaces). We only have to deal with remaining SP|HT |
304 | return c == ' ' || c == '\t'; |
305 | } |
306 | |
307 | static bool isTEXT(char c) |
308 | { |
309 | // TEXT = <any OCTET except CTLs, |
310 | // but including LWS> |
311 | return !isCTL(c) || isLWS(c); |
312 | } |
313 | |
314 | static bool isSeparator(char c) |
315 | { |
316 | // separators = "(" | ")" | "<" | ">" | "@" |
317 | // | "," | ";" | ":" | "\" | <"> |
318 | // | "/" | "[" | "]" | "?" | "=" |
319 | // | "{" | "}" | SP | HT |
320 | static const char separators[] = "()<>@,;:\\\"/[]?={}" ; |
321 | static const char *end = separators + sizeof separators - 1; |
322 | return isLWS(c) || std::find(separators, end, c) != end; |
323 | } |
324 | |
325 | static QByteArray unescapeMaxAge(const QByteArray &value) |
326 | { |
327 | if (value.size() < 2 || value[0] != '"') |
328 | return value; |
329 | |
330 | Q_ASSERT(value[value.size() - 1] == '"'); |
331 | return value.mid(1, value.size() - 2); |
332 | } |
333 | |
334 | static bool isTOKEN(char c) |
335 | { |
336 | // token = 1*<any CHAR except CTLs or separators> |
337 | return isCHAR(c) && !isCTL(c) && !isSeparator(c); |
338 | } |
339 | |
340 | /* |
341 | |
342 | RFC6797, 6.1 Strict-Transport-Security HTTP Response Header Field. |
343 | Syntax: |
344 | |
345 | Strict-Tranposrt-Security = "Strict-Transport-Security" ":" |
346 | [ directive ] *( ";" [ directive ] ) |
347 | |
348 | directive = directive-name [ "=" directive-value ] |
349 | directive-name = token |
350 | directive-value = token | quoted-string |
351 | |
352 | RFC 2616, 2.2 Basic Rules. |
353 | |
354 | token = 1*<any CHAR except CTLs or separators> |
355 | quoted-string = ( <"> *(qdtext | quoted-pair ) <"> ) |
356 | |
357 | |
358 | qdtext = <any TEXT except <">> |
359 | quoted-pair = "\" CHAR |
360 | |
361 | */ |
362 | |
363 | bool QHstsHeaderParser::(const QList<QPair<QByteArray, QByteArray>> &) |
364 | { |
365 | for (const auto &h : headers) { |
366 | // We use '==' since header name was already 'trimmed' for us: |
367 | if (h.first == "Strict-Transport-Security" ) { |
368 | header = h.second; |
369 | // RFC6797, 8.1: |
370 | // |
371 | // The UA MUST ignore any STS header fields not conforming to the |
372 | // grammar specified in Section 6.1 ("Strict-Transport-Security HTTP |
373 | // Response Header Field"). |
374 | // |
375 | // If a UA receives more than one STS header field in an HTTP |
376 | // response message over secure transport, then the UA MUST process |
377 | // only the first such header field. |
378 | // |
379 | // We read this as: ignore all invalid headers and take the first valid: |
380 | if (parseSTSHeader() && maxAgeFound) { |
381 | expiry = QDateTime::currentDateTimeUtc().addSecs(maxAge); |
382 | return true; |
383 | } |
384 | } |
385 | } |
386 | |
387 | // In case it was set by a syntactically correct header (but without |
388 | // REQUIRED max-age directive): |
389 | subDomainsFound = false; |
390 | |
391 | return false; |
392 | } |
393 | |
394 | bool QHstsHeaderParser::() |
395 | { |
396 | expiry = QDateTime(); |
397 | maxAgeFound = false; |
398 | subDomainsFound = false; |
399 | maxAge = 0; |
400 | tokenPos = 0; |
401 | token.clear(); |
402 | |
403 | while (tokenPos < header.size()) { |
404 | if (!parseDirective()) |
405 | return false; |
406 | |
407 | if (token.size() && token != ";" ) { |
408 | // After a directive we can only have a ";" or no more tokens. |
409 | // Invalid syntax. |
410 | return false; |
411 | } |
412 | } |
413 | |
414 | return true; |
415 | } |
416 | |
417 | bool QHstsHeaderParser::() |
418 | { |
419 | // RFC 6797, 6.1: |
420 | // |
421 | // directive = directive-name [ "=" directive-value ] |
422 | // directive-name = token |
423 | // directive-value = token | quoted-string |
424 | |
425 | |
426 | // RFC 2616, 2.2: |
427 | // |
428 | // token = 1*<any CHAR except CTLs or separators> |
429 | |
430 | if (!nextToken()) |
431 | return false; |
432 | |
433 | if (!token.size()) // No more data, but no error. |
434 | return true; |
435 | |
436 | if (token == ";" ) // That's a weird grammar, but that's what it is. |
437 | return true; |
438 | |
439 | if (!isTOKEN(token[0])) // Not a valid directive-name. |
440 | return false; |
441 | |
442 | const QByteArray directiveName = token; |
443 | // 2. Try to read "=" or ";". |
444 | if (!nextToken()) |
445 | return false; |
446 | |
447 | QByteArray directiveValue; |
448 | if (token == ";" ) // No directive-value |
449 | return processDirective(directiveName, directiveValue); |
450 | |
451 | if (token == "=" ) { |
452 | // We expect a directive-value now: |
453 | if (!nextToken() || !token.size()) |
454 | return false; |
455 | directiveValue = token; |
456 | } else if (token.size()) { |
457 | // Invalid syntax: |
458 | return false; |
459 | } |
460 | |
461 | if (!processDirective(directiveName, directiveValue)) |
462 | return false; |
463 | |
464 | // Read either ";", or 'end of header', or some invalid token. |
465 | return nextToken(); |
466 | } |
467 | |
468 | bool QHstsHeaderParser::(const QByteArray &name, const QByteArray &value) |
469 | { |
470 | Q_ASSERT(name.size()); |
471 | // RFC6797 6.1/3 Directive names are case-insensitive |
472 | if (name.compare("max-age" , Qt::CaseInsensitive) == 0) { |
473 | // RFC 6797, 6.1.1 |
474 | // The syntax of the max-age directive's REQUIRED value (after |
475 | // quoted-string unescaping, if necessary) is defined as: |
476 | // |
477 | // max-age-value = delta-seconds |
478 | if (maxAgeFound) { |
479 | // RFC 6797, 6.1/2: |
480 | // All directives MUST appear only once in an STS header field. |
481 | return false; |
482 | } |
483 | |
484 | const QByteArray unescapedValue = unescapeMaxAge(value); |
485 | if (!unescapedValue.size()) |
486 | return false; |
487 | |
488 | bool ok = false; |
489 | const qint64 age = unescapedValue.toLongLong(&ok); |
490 | if (!ok || age < 0) |
491 | return false; |
492 | |
493 | maxAge = age; |
494 | maxAgeFound = true; |
495 | } else if (name.compare("includesubdomains" , Qt::CaseInsensitive) == 0) { |
496 | // RFC 6797, 6.1.2. The includeSubDomains Directive. |
497 | // The OPTIONAL "includeSubDomains" directive is a valueless directive. |
498 | |
499 | if (subDomainsFound) { |
500 | // RFC 6797, 6.1/2: |
501 | // All directives MUST appear only once in an STS header field. |
502 | return false; |
503 | } |
504 | |
505 | subDomainsFound = true; |
506 | } // else we do nothing, skip unknown directives (RFC 6797, 6.1/5) |
507 | |
508 | return true; |
509 | } |
510 | |
511 | bool QHstsHeaderParser::() |
512 | { |
513 | // Returns true if we found a valid token or we have no more data (token is |
514 | // empty then). |
515 | |
516 | token.clear(); |
517 | |
518 | // Fortunately enough, by this point qhttpnetworkreply already got rid of |
519 | // [CRLF] parts, but we can have 1*(SP|HT) yet. |
520 | while (tokenPos < header.size() && isLWS(header[tokenPos])) |
521 | ++tokenPos; |
522 | |
523 | if (tokenPos == header.size()) |
524 | return true; |
525 | |
526 | const char ch = header[tokenPos]; |
527 | if (ch == ';' || ch == '=') { |
528 | token.append(ch); |
529 | ++tokenPos; |
530 | return true; |
531 | } |
532 | |
533 | // RFC 2616, 2.2. |
534 | // |
535 | // quoted-string = ( <"> *(qdtext | quoted-pair ) <"> ) |
536 | // qdtext = <any TEXT except <">> |
537 | if (ch == '"') { |
538 | int last = tokenPos + 1; |
539 | while (last < header.size()) { |
540 | if (header[last] == '"') { |
541 | // The end of a quoted-string. |
542 | break; |
543 | } else if (header[last] == '\\') { |
544 | // quoted-pair = "\" CHAR |
545 | if (last + 1 < header.size() && isCHAR(header[last + 1])) |
546 | last += 2; |
547 | else |
548 | return false; |
549 | } else { |
550 | if (!isTEXT(header[last])) |
551 | return false; |
552 | ++last; |
553 | } |
554 | } |
555 | |
556 | if (last >= header.size()) // no closing '"': |
557 | return false; |
558 | |
559 | token = header.mid(tokenPos, last - tokenPos + 1); |
560 | tokenPos = last + 1; |
561 | return true; |
562 | } |
563 | |
564 | // RFC 2616, 2.2: |
565 | // |
566 | // token = 1*<any CHAR except CTLs or separators> |
567 | if (!isTOKEN(ch)) |
568 | return false; |
569 | |
570 | int last = tokenPos + 1; |
571 | while (last < header.size() && isTOKEN(header[last])) |
572 | ++last; |
573 | |
574 | token = header.mid(tokenPos, last - tokenPos); |
575 | tokenPos = last; |
576 | |
577 | return true; |
578 | } |
579 | |
580 | QT_END_NAMESPACE |
581 | |