ResourceLoader.cpp 21 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572
  1. /*
  2. * Copyright (c) 2018-2023, Andreas Kling <kling@serenityos.org>
  3. * Copyright (c) 2022, Dex♪ <dexes.ttp@gmail.com>
  4. *
  5. * SPDX-License-Identifier: BSD-2-Clause
  6. */
  7. #include <AK/Debug.h>
  8. #include <LibCore/DateTime.h>
  9. #include <LibCore/Directory.h>
  10. #include <LibCore/ElapsedTimer.h>
  11. #include <LibCore/MimeData.h>
  12. #include <LibCore/Resource.h>
  13. #include <LibWeb/Cookie/Cookie.h>
  14. #include <LibWeb/Cookie/ParsedCookie.h>
  15. #include <LibWeb/Fetch/Infrastructure/URL.h>
  16. #include <LibWeb/Loader/ContentFilter.h>
  17. #include <LibWeb/Loader/GeneratedPagesLoader.h>
  18. #include <LibWeb/Loader/LoadRequest.h>
  19. #include <LibWeb/Loader/ProxyMappings.h>
  20. #include <LibWeb/Loader/Resource.h>
  21. #include <LibWeb/Loader/ResourceLoader.h>
  22. #include <LibWeb/Loader/UserAgent.h>
  23. #include <LibWeb/Platform/EventLoopPlugin.h>
  24. #include <LibWeb/Platform/Timer.h>
  25. namespace Web {
  26. ResourceLoaderConnectorRequest::ResourceLoaderConnectorRequest() = default;
  27. ResourceLoaderConnectorRequest::~ResourceLoaderConnectorRequest() = default;
  28. ResourceLoaderConnector::ResourceLoaderConnector() = default;
  29. ResourceLoaderConnector::~ResourceLoaderConnector() = default;
  30. static RefPtr<ResourceLoader> s_resource_loader;
  31. void ResourceLoader::initialize(RefPtr<ResourceLoaderConnector> connector)
  32. {
  33. if (connector)
  34. s_resource_loader = ResourceLoader::try_create(connector.release_nonnull()).release_value_but_fixme_should_propagate_errors();
  35. }
  36. ResourceLoader& ResourceLoader::the()
  37. {
  38. if (!s_resource_loader) {
  39. dbgln("Web::ResourceLoader was not initialized");
  40. VERIFY_NOT_REACHED();
  41. }
  42. return *s_resource_loader;
  43. }
  44. ErrorOr<NonnullRefPtr<ResourceLoader>> ResourceLoader::try_create(NonnullRefPtr<ResourceLoaderConnector> connector)
  45. {
  46. return adopt_nonnull_ref_or_enomem(new (nothrow) ResourceLoader(move(connector)));
  47. }
  48. ResourceLoader::ResourceLoader(NonnullRefPtr<ResourceLoaderConnector> connector)
  49. : m_connector(move(connector))
  50. , m_user_agent(MUST(String::from_utf8(default_user_agent)))
  51. , m_platform(MUST(String::from_utf8(default_platform)))
  52. {
  53. }
  54. void ResourceLoader::prefetch_dns(URL::URL const& url)
  55. {
  56. if (url.scheme().is_one_of("file"sv, "data"sv))
  57. return;
  58. if (ContentFilter::the().is_filtered(url)) {
  59. dbgln("ResourceLoader: Refusing to prefetch DNS for '{}': \033[31;1mURL was filtered\033[0m", url);
  60. return;
  61. }
  62. m_connector->prefetch_dns(url);
  63. }
  64. void ResourceLoader::preconnect(URL::URL const& url)
  65. {
  66. if (url.scheme().is_one_of("file"sv, "data"sv))
  67. return;
  68. if (ContentFilter::the().is_filtered(url)) {
  69. dbgln("ResourceLoader: Refusing to pre-connect to '{}': \033[31;1mURL was filtered\033[0m", url);
  70. return;
  71. }
  72. m_connector->preconnect(url);
  73. }
  74. static HashMap<LoadRequest, NonnullRefPtr<Resource>> s_resource_cache;
  75. RefPtr<Resource> ResourceLoader::load_resource(Resource::Type type, LoadRequest& request)
  76. {
  77. if (!request.is_valid())
  78. return nullptr;
  79. bool use_cache = request.url().scheme() != "file";
  80. if (use_cache) {
  81. auto it = s_resource_cache.find(request);
  82. if (it != s_resource_cache.end()) {
  83. if (it->value->type() != type) {
  84. dbgln("FIXME: Not using cached resource for {} since there's a type mismatch.", request.url());
  85. } else {
  86. dbgln_if(CACHE_DEBUG, "Reusing cached resource for: {}", request.url());
  87. return it->value;
  88. }
  89. }
  90. }
  91. auto resource = Resource::create({}, type, request);
  92. if (use_cache)
  93. s_resource_cache.set(request, resource);
  94. load(
  95. request,
  96. [=](auto data, auto& headers, auto status_code) {
  97. const_cast<Resource&>(*resource).did_load({}, data, headers, status_code);
  98. },
  99. [=](auto& error, auto status_code, auto, auto) {
  100. const_cast<Resource&>(*resource).did_fail({}, error, status_code);
  101. });
  102. return resource;
  103. }
  104. static ByteString sanitized_url_for_logging(URL::URL const& url)
  105. {
  106. if (url.scheme() == "data"sv)
  107. return "[data URL]"sv;
  108. return url.to_byte_string();
  109. }
  110. static void emit_signpost(ByteString const& message, int id)
  111. {
  112. (void)message;
  113. (void)id;
  114. }
  115. static void store_response_cookies(Page& page, URL::URL const& url, ByteString const& set_cookie_entry)
  116. {
  117. auto cookie = Cookie::parse_cookie(set_cookie_entry);
  118. if (!cookie.has_value())
  119. return;
  120. page.client().page_did_set_cookie(url, cookie.value(), Cookie::Source::Http); // FIXME: Determine cookie source correctly
  121. }
  122. static HTTP::HeaderMap response_headers_for_file(StringView path, Optional<time_t> const& modified_time)
  123. {
  124. // For file:// and resource:// URLs, we have to guess the MIME type, since there's no HTTP header to tell us what
  125. // it is. We insert a fake Content-Type header here, so that clients can use it to learn the MIME type.
  126. auto mime_type = Core::guess_mime_type_based_on_filename(path);
  127. HTTP::HeaderMap response_headers;
  128. response_headers.set("Content-Type"sv, mime_type);
  129. if (modified_time.has_value()) {
  130. auto const datetime = Core::DateTime::from_timestamp(modified_time.value());
  131. response_headers.set("Last-Modified"sv, datetime.to_byte_string("%a, %d %b %Y %H:%M:%S GMT"sv, Core::DateTime::LocalTime::No));
  132. }
  133. return response_headers;
  134. }
  135. static void log_request_start(LoadRequest const& request)
  136. {
  137. auto url_for_logging = sanitized_url_for_logging(request.url());
  138. emit_signpost(ByteString::formatted("Starting load: {}", url_for_logging), request.id());
  139. dbgln_if(SPAM_DEBUG, "ResourceLoader: Starting load of: \"{}\"", url_for_logging);
  140. }
  141. static void log_success(LoadRequest const& request)
  142. {
  143. auto url_for_logging = sanitized_url_for_logging(request.url());
  144. auto load_time_ms = request.load_time().to_milliseconds();
  145. emit_signpost(ByteString::formatted("Finished load: {}", url_for_logging), request.id());
  146. dbgln_if(SPAM_DEBUG, "ResourceLoader: Finished load of: \"{}\", Duration: {}ms", url_for_logging, load_time_ms);
  147. }
  148. template<typename ErrorType>
  149. static void log_failure(LoadRequest const& request, ErrorType const& error)
  150. {
  151. auto url_for_logging = sanitized_url_for_logging(request.url());
  152. auto load_time_ms = request.load_time().to_milliseconds();
  153. emit_signpost(ByteString::formatted("Failed load: {}", url_for_logging), request.id());
  154. dbgln("ResourceLoader: Failed load of: \"{}\", \033[31;1mError: {}\033[0m, Duration: {}ms", url_for_logging, error, load_time_ms);
  155. }
  156. static void log_filtered_request(LoadRequest const& request)
  157. {
  158. auto url_for_logging = sanitized_url_for_logging(request.url());
  159. dbgln("ResourceLoader: Filtered request to: \"{}\"", url_for_logging);
  160. }
  161. static bool should_block_request(LoadRequest const& request)
  162. {
  163. auto const& url = request.url();
  164. auto is_port_blocked = [](int port) {
  165. static constexpr auto ports = to_array({ 1, 7, 9, 11, 13, 15, 17, 19, 20, 21, 22, 23, 25, 37, 42,
  166. 43, 53, 77, 79, 87, 95, 101, 102, 103, 104, 109, 110, 111, 113, 115, 117, 119, 123, 135, 139,
  167. 143, 179, 389, 465, 512, 513, 514, 515, 526, 530, 531, 532, 540, 556, 563, 587, 601, 636,
  168. 993, 995, 2049, 3659, 4045, 6000, 6379, 6665, 6666, 6667, 6668, 6669 });
  169. return ports.first_index_of(port).has_value();
  170. };
  171. if (is_port_blocked(url.port_or_default())) {
  172. log_failure(request, ByteString::formatted("Port #{} is blocked", url.port_or_default()));
  173. return true;
  174. }
  175. if (ContentFilter::the().is_filtered(url)) {
  176. log_filtered_request(request);
  177. return true;
  178. }
  179. return false;
  180. }
  181. void ResourceLoader::load(LoadRequest& request, SuccessCallback success_callback, ErrorCallback error_callback, Optional<u32> timeout, TimeoutCallback timeout_callback)
  182. {
  183. auto const& url = request.url();
  184. log_request_start(request);
  185. request.start_timer();
  186. if (should_block_request(request)) {
  187. error_callback("Request was blocked", {}, {}, {});
  188. return;
  189. }
  190. auto respond_directory_page = [](LoadRequest const& request, URL::URL const& url, SuccessCallback const& success_callback, ErrorCallback const& error_callback) {
  191. auto maybe_response = load_file_directory_page(url);
  192. if (maybe_response.is_error()) {
  193. log_failure(request, maybe_response.error());
  194. if (error_callback)
  195. error_callback(ByteString::formatted("{}", maybe_response.error()), 500u, {}, {});
  196. return;
  197. }
  198. log_success(request);
  199. HTTP::HeaderMap response_headers;
  200. response_headers.set("Content-Type"sv, "text/html"sv);
  201. success_callback(maybe_response.release_value().bytes(), response_headers, {});
  202. };
  203. if (url.scheme() == "about") {
  204. dbgln_if(SPAM_DEBUG, "Loading about: URL {}", url);
  205. log_success(request);
  206. HTTP::HeaderMap response_headers;
  207. response_headers.set("Content-Type", "text/html; charset=UTF-8");
  208. // About version page
  209. if (url.path_segment_at_index(0) == "version") {
  210. success_callback(MUST(load_about_version_page()).bytes(), response_headers, {});
  211. return;
  212. }
  213. // Other about static HTML pages
  214. auto resource = Core::Resource::load_from_uri(MUST(String::formatted("resource://ladybird/{}.html", url.path_segment_at_index(0))));
  215. if (!resource.is_error()) {
  216. auto data = resource.value()->data();
  217. success_callback(data, response_headers, {});
  218. return;
  219. }
  220. Platform::EventLoopPlugin::the().deferred_invoke([success_callback = move(success_callback), response_headers = move(response_headers)] {
  221. success_callback(ByteString::empty().to_byte_buffer(), response_headers, {});
  222. });
  223. return;
  224. }
  225. if (url.scheme() == "data") {
  226. auto data_url_or_error = Fetch::Infrastructure::process_data_url(url);
  227. if (data_url_or_error.is_error()) {
  228. auto error_message = data_url_or_error.error().string_literal();
  229. log_failure(request, error_message);
  230. error_callback(error_message, {}, {}, {});
  231. return;
  232. }
  233. auto data_url = data_url_or_error.release_value();
  234. dbgln_if(SPAM_DEBUG, "ResourceLoader loading a data URL with mime-type: '{}', payload='{}'",
  235. MUST(data_url.mime_type.serialized()),
  236. StringView(data_url.body.bytes()));
  237. HTTP::HeaderMap response_headers;
  238. response_headers.set("Content-Type", MUST(data_url.mime_type.serialized()).to_byte_string());
  239. log_success(request);
  240. Platform::EventLoopPlugin::the().deferred_invoke([data = move(data_url.body), response_headers = move(response_headers), success_callback = move(success_callback)] {
  241. success_callback(data, response_headers, {});
  242. });
  243. return;
  244. }
  245. if (url.scheme() == "resource") {
  246. auto resource = Core::Resource::load_from_uri(url.serialize());
  247. if (resource.is_error()) {
  248. log_failure(request, resource.error());
  249. return;
  250. }
  251. // When resource URI is a directory use file directory loader to generate response
  252. if (resource.value()->is_directory()) {
  253. respond_directory_page(request, resource.value()->file_url(), success_callback, error_callback);
  254. return;
  255. }
  256. auto data = resource.value()->data();
  257. auto response_headers = response_headers_for_file(url.serialize_path(), resource.value()->modified_time());
  258. log_success(request);
  259. success_callback(data, response_headers, {});
  260. return;
  261. }
  262. if (url.scheme() == "file") {
  263. if (request.page())
  264. m_page = request.page();
  265. if (!m_page.has_value()) {
  266. log_failure(request, "INTERNAL ERROR: No Page for request");
  267. return;
  268. }
  269. FileRequest file_request(url.serialize_path(), [this, success_callback = move(success_callback), error_callback = move(error_callback), request, respond_directory_page](ErrorOr<i32> file_or_error) {
  270. --m_pending_loads;
  271. if (on_load_counter_change)
  272. on_load_counter_change();
  273. if (file_or_error.is_error()) {
  274. log_failure(request, file_or_error.error());
  275. if (error_callback) {
  276. auto status = file_or_error.error().code() == ENOENT ? 404u : 500u;
  277. error_callback(ByteString::formatted("{}", file_or_error.error()), status, {}, {});
  278. }
  279. return;
  280. }
  281. auto const fd = file_or_error.value();
  282. // When local file is a directory use file directory loader to generate response
  283. auto maybe_is_valid_directory = Core::Directory::is_valid_directory(fd);
  284. if (!maybe_is_valid_directory.is_error() && maybe_is_valid_directory.value()) {
  285. respond_directory_page(request, request.url(), success_callback, error_callback);
  286. return;
  287. }
  288. auto st_or_error = Core::System::fstat(fd);
  289. if (st_or_error.is_error()) {
  290. log_failure(request, st_or_error.error());
  291. if (error_callback)
  292. error_callback(ByteString::formatted("{}", st_or_error.error()), 500u, {}, {});
  293. return;
  294. }
  295. // Try to read file normally
  296. auto maybe_file = Core::File::adopt_fd(fd, Core::File::OpenMode::Read);
  297. if (maybe_file.is_error()) {
  298. log_failure(request, maybe_file.error());
  299. if (error_callback)
  300. error_callback(ByteString::formatted("{}", maybe_file.error()), 500u, {}, {});
  301. return;
  302. }
  303. auto file = maybe_file.release_value();
  304. auto maybe_data = file->read_until_eof();
  305. if (maybe_data.is_error()) {
  306. log_failure(request, maybe_data.error());
  307. if (error_callback)
  308. error_callback(ByteString::formatted("{}", maybe_data.error()), 500u, {}, {});
  309. return;
  310. }
  311. auto data = maybe_data.release_value();
  312. auto response_headers = response_headers_for_file(request.url().serialize_path(), st_or_error.value().st_mtime);
  313. log_success(request);
  314. success_callback(data, response_headers, {});
  315. });
  316. (*m_page)->client().request_file(move(file_request));
  317. ++m_pending_loads;
  318. if (on_load_counter_change)
  319. on_load_counter_change();
  320. return;
  321. }
  322. if (url.scheme() == "http" || url.scheme() == "https") {
  323. auto protocol_request = start_network_request(request);
  324. if (!protocol_request) {
  325. if (error_callback)
  326. error_callback("Failed to start network request"sv, {}, {}, {});
  327. return;
  328. }
  329. if (timeout.has_value() && timeout.value() > 0) {
  330. auto timer = Platform::Timer::create_single_shot(timeout.value(), nullptr);
  331. timer->on_timeout = [timer, protocol_request, timeout_callback = move(timeout_callback)] {
  332. protocol_request->stop();
  333. if (timeout_callback)
  334. timeout_callback();
  335. };
  336. timer->start();
  337. }
  338. auto on_buffered_request_finished = [this, success_callback = move(success_callback), error_callback = move(error_callback), request, &protocol_request = *protocol_request](bool success, auto, auto& response_headers, auto status_code, ReadonlyBytes payload) mutable {
  339. handle_network_response_headers(request, response_headers);
  340. finish_network_request(protocol_request);
  341. if (!success || (status_code.has_value() && *status_code >= 400 && *status_code <= 599 && (payload.is_empty() || !request.is_main_resource()))) {
  342. StringBuilder error_builder;
  343. if (status_code.has_value())
  344. error_builder.appendff("Load failed: {}", *status_code);
  345. else
  346. error_builder.append("Load failed"sv);
  347. log_failure(request, error_builder.string_view());
  348. if (error_callback)
  349. error_callback(error_builder.to_byte_string(), status_code, payload, response_headers);
  350. return;
  351. }
  352. log_success(request);
  353. success_callback(payload, response_headers, status_code);
  354. };
  355. protocol_request->set_buffered_request_finished_callback(move(on_buffered_request_finished));
  356. return;
  357. }
  358. auto not_implemented_error = ByteString::formatted("Protocol not implemented: {}", url.scheme());
  359. log_failure(request, not_implemented_error);
  360. if (error_callback)
  361. error_callback(not_implemented_error, {}, {}, {});
  362. }
  363. void ResourceLoader::load_unbuffered(LoadRequest& request, OnHeadersReceived on_headers_received, OnDataReceived on_data_received, OnComplete on_complete)
  364. {
  365. auto const& url = request.url();
  366. log_request_start(request);
  367. request.start_timer();
  368. if (should_block_request(request)) {
  369. on_complete(false, "Request was blocked"sv);
  370. return;
  371. }
  372. if (!url.scheme().is_one_of("http"sv, "https"sv)) {
  373. // FIXME: Non-network requests from fetch should not go through this path.
  374. on_complete(false, "Cannot establish connection non-network scheme"sv);
  375. return;
  376. }
  377. auto protocol_request = start_network_request(request);
  378. if (!protocol_request) {
  379. on_complete(false, "Failed to start network request"sv);
  380. return;
  381. }
  382. auto protocol_headers_received = [this, on_headers_received = move(on_headers_received), request](auto const& response_headers, auto status_code) {
  383. handle_network_response_headers(request, response_headers);
  384. on_headers_received(response_headers, move(status_code));
  385. };
  386. auto protocol_data_received = [on_data_received = move(on_data_received)](auto data) {
  387. on_data_received(data);
  388. };
  389. auto protocol_complete = [this, on_complete = move(on_complete), request, &protocol_request = *protocol_request](bool success, u64) {
  390. finish_network_request(protocol_request);
  391. if (success) {
  392. log_success(request);
  393. on_complete(true, {});
  394. } else {
  395. log_failure(request, "Request finished with error"sv);
  396. on_complete(false, "Request finished with error"sv);
  397. }
  398. };
  399. protocol_request->set_unbuffered_request_callbacks(move(protocol_headers_received), move(protocol_data_received), move(protocol_complete));
  400. }
  401. RefPtr<ResourceLoaderConnectorRequest> ResourceLoader::start_network_request(LoadRequest const& request)
  402. {
  403. auto proxy = ProxyMappings::the().proxy_for_url(request.url());
  404. HTTP::HeaderMap headers;
  405. for (auto const& it : request.headers()) {
  406. headers.set(it.key, it.value);
  407. }
  408. if (!headers.contains("User-Agent"))
  409. headers.set("User-Agent", m_user_agent.to_byte_string());
  410. if (!headers.contains("Accept-Encoding"))
  411. headers.set("Accept-Encoding", "gzip, deflate, br");
  412. auto protocol_request = m_connector->start_request(request.method(), request.url(), headers, request.body(), proxy);
  413. if (!protocol_request) {
  414. log_failure(request, "Failed to initiate load"sv);
  415. return nullptr;
  416. }
  417. protocol_request->on_certificate_requested = []() -> ResourceLoaderConnectorRequest::CertificateAndKey {
  418. return {};
  419. };
  420. ++m_pending_loads;
  421. if (on_load_counter_change)
  422. on_load_counter_change();
  423. m_active_requests.set(*protocol_request);
  424. return protocol_request;
  425. }
  426. void ResourceLoader::handle_network_response_headers(LoadRequest const& request, HTTP::HeaderMap const& response_headers)
  427. {
  428. if (!request.page())
  429. return;
  430. for (auto const& [header, value] : response_headers.headers()) {
  431. if (header.equals_ignoring_ascii_case("Set-Cookie"sv)) {
  432. store_response_cookies(*request.page(), request.url(), value);
  433. }
  434. }
  435. if (auto cache_control = response_headers.get("Cache-Control"); cache_control.has_value()) {
  436. if (cache_control.value().contains("no-store"sv))
  437. s_resource_cache.remove(request);
  438. }
  439. }
  440. void ResourceLoader::finish_network_request(NonnullRefPtr<ResourceLoaderConnectorRequest> const& protocol_request)
  441. {
  442. --m_pending_loads;
  443. if (on_load_counter_change)
  444. on_load_counter_change();
  445. Platform::EventLoopPlugin::the().deferred_invoke([this, protocol_request] {
  446. m_active_requests.remove(protocol_request);
  447. });
  448. }
  449. void ResourceLoader::clear_cache()
  450. {
  451. dbgln_if(CACHE_DEBUG, "Clearing {} items from ResourceLoader cache", s_resource_cache.size());
  452. s_resource_cache.clear();
  453. }
  454. void ResourceLoader::evict_from_cache(LoadRequest const& request)
  455. {
  456. dbgln_if(CACHE_DEBUG, "Removing resource {} from cache", request.url());
  457. s_resource_cache.remove(request);
  458. }
  459. }