ResourceLoader.cpp 14 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367
  1. /*
  2. * Copyright (c) 2018-2022, 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/Base64.h>
  8. #include <AK/Debug.h>
  9. #include <AK/JsonObject.h>
  10. #include <LibCore/ElapsedTimer.h>
  11. #include <LibCore/File.h>
  12. #include <LibWeb/Loader/ContentFilter.h>
  13. #include <LibWeb/Loader/LoadRequest.h>
  14. #include <LibWeb/Loader/ProxyMappings.h>
  15. #include <LibWeb/Loader/Resource.h>
  16. #include <LibWeb/Loader/ResourceLoader.h>
  17. #include <LibWeb/Platform/EventLoopPlugin.h>
  18. #include <LibWeb/Platform/Timer.h>
  19. #ifdef __serenity__
  20. # include <serenity.h>
  21. #endif
  22. namespace Web {
  23. ResourceLoaderConnectorRequest::ResourceLoaderConnectorRequest() = default;
  24. ResourceLoaderConnectorRequest::~ResourceLoaderConnectorRequest() = default;
  25. ResourceLoaderConnector::ResourceLoaderConnector() = default;
  26. ResourceLoaderConnector::~ResourceLoaderConnector() = default;
  27. static RefPtr<ResourceLoader> s_resource_loader;
  28. void ResourceLoader::initialize(RefPtr<ResourceLoaderConnector> connector)
  29. {
  30. if (connector)
  31. s_resource_loader = ResourceLoader::try_create(connector.release_nonnull()).release_value_but_fixme_should_propagate_errors();
  32. }
  33. ResourceLoader& ResourceLoader::the()
  34. {
  35. if (!s_resource_loader) {
  36. dbgln("Web::ResourceLoader was not initialized");
  37. VERIFY_NOT_REACHED();
  38. }
  39. return *s_resource_loader;
  40. }
  41. ErrorOr<NonnullRefPtr<ResourceLoader>> ResourceLoader::try_create(NonnullRefPtr<ResourceLoaderConnector> connector)
  42. {
  43. return adopt_nonnull_ref_or_enomem(new (nothrow) ResourceLoader(move(connector)));
  44. }
  45. ResourceLoader::ResourceLoader(NonnullRefPtr<ResourceLoaderConnector> connector)
  46. : m_connector(move(connector))
  47. , m_user_agent(default_user_agent)
  48. {
  49. }
  50. void ResourceLoader::prefetch_dns(AK::URL const& url)
  51. {
  52. if (ContentFilter::the().is_filtered(url)) {
  53. dbgln("ResourceLoader: Refusing to prefetch DNS for '{}': \033[31;1mURL was filtered\033[0m", url);
  54. return;
  55. }
  56. m_connector->prefetch_dns(url);
  57. }
  58. void ResourceLoader::preconnect(AK::URL const& url)
  59. {
  60. if (ContentFilter::the().is_filtered(url)) {
  61. dbgln("ResourceLoader: Refusing to pre-connect to '{}': \033[31;1mURL was filtered\033[0m", url);
  62. return;
  63. }
  64. m_connector->preconnect(url);
  65. }
  66. static HashMap<LoadRequest, NonnullRefPtr<Resource>> s_resource_cache;
  67. RefPtr<Resource> ResourceLoader::load_resource(Resource::Type type, LoadRequest& request)
  68. {
  69. if (!request.is_valid())
  70. return nullptr;
  71. bool use_cache = request.url().scheme() != "file";
  72. if (use_cache) {
  73. auto it = s_resource_cache.find(request);
  74. if (it != s_resource_cache.end()) {
  75. if (it->value->type() != type) {
  76. dbgln("FIXME: Not using cached resource for {} since there's a type mismatch.", request.url());
  77. } else {
  78. dbgln_if(CACHE_DEBUG, "Reusing cached resource for: {}", request.url());
  79. return it->value;
  80. }
  81. }
  82. }
  83. auto resource = Resource::create({}, type, request);
  84. if (use_cache)
  85. s_resource_cache.set(request, resource);
  86. load(
  87. request,
  88. [=](auto data, auto& headers, auto status_code) {
  89. const_cast<Resource&>(*resource).did_load({}, data, headers, status_code);
  90. },
  91. [=](auto& error, auto status_code) {
  92. const_cast<Resource&>(*resource).did_fail({}, error, status_code);
  93. });
  94. return resource;
  95. }
  96. static String sanitized_url_for_logging(AK::URL const& url)
  97. {
  98. if (url.scheme() == "data"sv)
  99. return String::formatted("[data URL, mime-type={}, size={}]", url.data_mime_type(), url.data_payload().length());
  100. return url.to_string();
  101. }
  102. static void emit_signpost(String const& message, int id)
  103. {
  104. #ifdef __serenity__
  105. auto string_id = perf_register_string(message.characters(), message.length());
  106. perf_event(PERF_EVENT_SIGNPOST, string_id, id);
  107. #else
  108. (void)message;
  109. (void)id;
  110. #endif
  111. }
  112. static size_t resource_id = 0;
  113. void ResourceLoader::load(LoadRequest& request, Function<void(ReadonlyBytes, HashMap<String, String, CaseInsensitiveStringTraits> const& response_headers, Optional<u32> status_code)> success_callback, Function<void(String const&, Optional<u32> status_code)> error_callback, Optional<u32> timeout, Function<void()> timeout_callback)
  114. {
  115. auto& url = request.url();
  116. request.start_timer();
  117. auto id = resource_id++;
  118. auto url_for_logging = sanitized_url_for_logging(url);
  119. emit_signpost(String::formatted("Starting load: {}", url_for_logging), id);
  120. dbgln("ResourceLoader: Starting load of: \"{}\"", url_for_logging);
  121. auto const log_success = [url_for_logging, id](auto const& request) {
  122. auto load_time_ms = request.load_time().to_milliseconds();
  123. emit_signpost(String::formatted("Finished load: {}", url_for_logging), id);
  124. dbgln("ResourceLoader: Finished load of: \"{}\", Duration: {}ms", url_for_logging, load_time_ms);
  125. };
  126. auto const log_failure = [url_for_logging, id](auto const& request, auto const error_message) {
  127. auto load_time_ms = request.load_time().to_milliseconds();
  128. emit_signpost(String::formatted("Failed load: {}", url_for_logging), id);
  129. dbgln("ResourceLoader: Failed load of: \"{}\", \033[31;1mError: {}\033[0m, Duration: {}ms", url_for_logging, error_message, load_time_ms);
  130. };
  131. if (is_port_blocked(url.port_or_default())) {
  132. log_failure(request, String::formatted("The port #{} is blocked", url.port_or_default()));
  133. return;
  134. }
  135. if (ContentFilter::the().is_filtered(url)) {
  136. auto filter_message = "URL was filtered"sv;
  137. log_failure(request, filter_message);
  138. error_callback(filter_message, {});
  139. return;
  140. }
  141. if (url.scheme() == "about") {
  142. dbgln_if(SPAM_DEBUG, "Loading about: URL {}", url);
  143. log_success(request);
  144. HashMap<String, String, CaseInsensitiveStringTraits> response_headers;
  145. response_headers.set("Content-Type", "text/html; charset=UTF-8");
  146. Platform::EventLoopPlugin::the().deferred_invoke([success_callback = move(success_callback), response_headers = move(response_headers)] {
  147. success_callback(String::empty().to_byte_buffer(), response_headers, {});
  148. });
  149. return;
  150. }
  151. if (url.scheme() == "data") {
  152. dbgln_if(SPAM_DEBUG, "ResourceLoader loading a data URL with mime-type: '{}', base64={}, payload='{}'",
  153. url.data_mime_type(),
  154. url.data_payload_is_base64(),
  155. url.data_payload());
  156. ByteBuffer data;
  157. if (url.data_payload_is_base64()) {
  158. auto data_maybe = decode_base64(url.data_payload());
  159. if (data_maybe.is_error()) {
  160. auto error_message = data_maybe.error().string_literal();
  161. log_failure(request, error_message);
  162. error_callback(error_message, {});
  163. return;
  164. }
  165. data = data_maybe.value();
  166. } else {
  167. data = url.data_payload().to_byte_buffer();
  168. }
  169. log_success(request);
  170. Platform::EventLoopPlugin::the().deferred_invoke([data = move(data), success_callback = move(success_callback)] {
  171. success_callback(data, {}, {});
  172. });
  173. return;
  174. }
  175. if (url.scheme() == "file") {
  176. if (request.page().has_value())
  177. m_page = request.page().value();
  178. if (!m_page.has_value())
  179. return;
  180. VERIFY(m_page.has_value());
  181. auto file_ref = make_ref_counted<FileRequest>(url.path());
  182. file_ref->on_file_request_finish = [this, success_callback = move(success_callback), error_callback = move(error_callback), log_success, log_failure, request, file_ref](ErrorOr<i32> file_or_error) {
  183. --m_pending_loads;
  184. if (on_load_counter_change)
  185. on_load_counter_change();
  186. if (file_or_error.is_error()) {
  187. log_failure(request, file_or_error.error());
  188. if (error_callback)
  189. error_callback(String::formatted("{}", file_or_error.error()), file_or_error.error().code());
  190. return;
  191. }
  192. auto const fd = file_or_error.value();
  193. auto maybe_file = Core::Stream::File::adopt_fd(fd, Core::Stream::OpenMode::Read);
  194. if (maybe_file.is_error()) {
  195. log_failure(request, maybe_file.error());
  196. if (error_callback)
  197. error_callback(String::formatted("{}", maybe_file.error()), maybe_file.error().code());
  198. return;
  199. }
  200. auto file = maybe_file.release_value();
  201. auto maybe_data = file->read_all();
  202. if (maybe_data.is_error()) {
  203. log_failure(request, maybe_data.error());
  204. if (error_callback)
  205. error_callback(String::formatted("{}", maybe_data.error()), maybe_data.error().code());
  206. return;
  207. }
  208. auto data = maybe_data.release_value();
  209. log_success(request);
  210. success_callback(data, {}, {});
  211. };
  212. m_page->client().request_file(file_ref);
  213. ++m_pending_loads;
  214. if (on_load_counter_change)
  215. on_load_counter_change();
  216. return;
  217. }
  218. if (url.scheme() == "http" || url.scheme() == "https" || url.scheme() == "gemini") {
  219. auto proxy = ProxyMappings::the().proxy_for_url(url);
  220. HashMap<String, String> headers;
  221. headers.set("User-Agent", m_user_agent);
  222. headers.set("Accept-Encoding", "gzip, deflate, br");
  223. for (auto& it : request.headers()) {
  224. headers.set(it.key, it.value);
  225. }
  226. auto protocol_request = m_connector->start_request(request.method(), url, headers, request.body(), proxy);
  227. if (!protocol_request) {
  228. auto start_request_failure_msg = "Failed to initiate load"sv;
  229. log_failure(request, start_request_failure_msg);
  230. if (error_callback)
  231. error_callback(start_request_failure_msg, {});
  232. return;
  233. }
  234. if (timeout.has_value() && timeout.value() > 0) {
  235. auto timer = Platform::Timer::create_single_shot(timeout.value(), nullptr);
  236. timer->on_timeout = [timer, protocol_request, timeout_callback = move(timeout_callback)]() mutable {
  237. protocol_request->stop();
  238. if (timeout_callback)
  239. timeout_callback();
  240. };
  241. timer->start();
  242. }
  243. m_active_requests.set(*protocol_request);
  244. protocol_request->on_buffered_request_finish = [this, success_callback = move(success_callback), error_callback = move(error_callback), log_success, log_failure, request, &protocol_request = *protocol_request](bool success, auto, auto& response_headers, auto status_code, ReadonlyBytes payload) {
  245. --m_pending_loads;
  246. if (on_load_counter_change)
  247. on_load_counter_change();
  248. if (!success || (status_code.has_value() && *status_code >= 400 && *status_code <= 599)) {
  249. StringBuilder error_builder;
  250. if (status_code.has_value())
  251. error_builder.appendff("Load failed: {}", *status_code);
  252. else
  253. error_builder.append("Load failed"sv);
  254. log_failure(request, error_builder.string_view());
  255. if (error_callback)
  256. error_callback(error_builder.to_string(), status_code);
  257. return;
  258. }
  259. log_success(request);
  260. success_callback(payload, response_headers, status_code);
  261. Platform::EventLoopPlugin::the().deferred_invoke([this, &protocol_request] {
  262. m_active_requests.remove(protocol_request);
  263. });
  264. };
  265. protocol_request->set_should_buffer_all_input(true);
  266. protocol_request->on_certificate_requested = []() -> ResourceLoaderConnectorRequest::CertificateAndKey {
  267. return {};
  268. };
  269. ++m_pending_loads;
  270. if (on_load_counter_change)
  271. on_load_counter_change();
  272. return;
  273. }
  274. auto not_implemented_error = String::formatted("Protocol not implemented: {}", url.scheme());
  275. log_failure(request, not_implemented_error);
  276. if (error_callback)
  277. error_callback(not_implemented_error, {});
  278. }
  279. void ResourceLoader::load(const AK::URL& url, Function<void(ReadonlyBytes, HashMap<String, String, CaseInsensitiveStringTraits> const& response_headers, Optional<u32> status_code)> success_callback, Function<void(String const&, Optional<u32> status_code)> error_callback, Optional<u32> timeout, Function<void()> timeout_callback)
  280. {
  281. LoadRequest request;
  282. request.set_url(url);
  283. load(request, move(success_callback), move(error_callback), timeout, move(timeout_callback));
  284. }
  285. bool ResourceLoader::is_port_blocked(int port)
  286. {
  287. int ports[] { 1, 7, 9, 11, 13, 15, 17, 19, 20, 21, 22, 23, 25, 37, 42,
  288. 43, 53, 77, 79, 87, 95, 101, 102, 103, 104, 109, 110, 111, 113,
  289. 115, 117, 119, 123, 135, 139, 143, 179, 389, 465, 512, 513, 514,
  290. 515, 526, 530, 531, 532, 540, 556, 563, 587, 601, 636, 993, 995,
  291. 2049, 3659, 4045, 6000, 6379, 6665, 6666, 6667, 6668, 6669, 9000 };
  292. for (auto blocked_port : ports)
  293. if (port == blocked_port)
  294. return true;
  295. return false;
  296. }
  297. void ResourceLoader::clear_cache()
  298. {
  299. dbgln_if(CACHE_DEBUG, "Clearing {} items from ResourceLoader cache", s_resource_cache.size());
  300. s_resource_cache.clear();
  301. }
  302. void ResourceLoader::evict_from_cache(LoadRequest const& request)
  303. {
  304. dbgln_if(CACHE_DEBUG, "Removing resource {} from cache", request.url());
  305. s_resource_cache.remove(request);
  306. }
  307. }