ResourceLoader.cpp 18 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472
  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 <AK/JsonObject.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/Platform/EventLoopPlugin.h>
  23. #include <LibWeb/Platform/Timer.h>
  24. #ifdef AK_OS_SERENITY
  25. # include <serenity.h>
  26. #endif
  27. namespace Web {
  28. ResourceLoaderConnectorRequest::ResourceLoaderConnectorRequest() = default;
  29. ResourceLoaderConnectorRequest::~ResourceLoaderConnectorRequest() = default;
  30. ResourceLoaderConnector::ResourceLoaderConnector() = default;
  31. ResourceLoaderConnector::~ResourceLoaderConnector() = default;
  32. static RefPtr<ResourceLoader> s_resource_loader;
  33. void ResourceLoader::initialize(RefPtr<ResourceLoaderConnector> connector)
  34. {
  35. if (connector)
  36. s_resource_loader = ResourceLoader::try_create(connector.release_nonnull()).release_value_but_fixme_should_propagate_errors();
  37. }
  38. ResourceLoader& ResourceLoader::the()
  39. {
  40. if (!s_resource_loader) {
  41. dbgln("Web::ResourceLoader was not initialized");
  42. VERIFY_NOT_REACHED();
  43. }
  44. return *s_resource_loader;
  45. }
  46. ErrorOr<NonnullRefPtr<ResourceLoader>> ResourceLoader::try_create(NonnullRefPtr<ResourceLoaderConnector> connector)
  47. {
  48. return adopt_nonnull_ref_or_enomem(new (nothrow) ResourceLoader(move(connector)));
  49. }
  50. ResourceLoader::ResourceLoader(NonnullRefPtr<ResourceLoaderConnector> connector)
  51. : m_connector(move(connector))
  52. , m_user_agent(MUST(String::from_utf8(default_user_agent)))
  53. , m_platform(MUST(String::from_utf8(default_platform)))
  54. {
  55. }
  56. void ResourceLoader::prefetch_dns(URL::URL const& url)
  57. {
  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. #ifdef AK_OS_SERENITY
  113. auto string_id = perf_register_string(message.characters(), message.length());
  114. perf_event(PERF_EVENT_SIGNPOST, string_id, id);
  115. #else
  116. (void)message;
  117. (void)id;
  118. #endif
  119. }
  120. static void store_response_cookies(Page& page, URL::URL const& url, ByteString const& cookies)
  121. {
  122. auto set_cookie_json_value = MUST(JsonValue::from_string(cookies));
  123. VERIFY(set_cookie_json_value.type() == JsonValue::Type::Array);
  124. for (auto const& set_cookie_entry : set_cookie_json_value.as_array().values()) {
  125. VERIFY(set_cookie_entry.type() == JsonValue::Type::String);
  126. auto cookie = Cookie::parse_cookie(set_cookie_entry.as_string());
  127. if (!cookie.has_value())
  128. continue;
  129. page.client().page_did_set_cookie(url, cookie.value(), Cookie::Source::Http); // FIXME: Determine cookie source correctly
  130. }
  131. }
  132. static size_t resource_id = 0;
  133. static HashMap<ByteString, ByteString, CaseInsensitiveStringTraits> response_headers_for_file(StringView path)
  134. {
  135. // For file:// and resource:// URLs, we have to guess the MIME type, since there's no HTTP header to tell us what
  136. // it is. We insert a fake Content-Type header here, so that clients can use it to learn the MIME type.
  137. auto mime_type = Core::guess_mime_type_based_on_filename(path);
  138. HashMap<ByteString, ByteString, CaseInsensitiveStringTraits> response_headers;
  139. response_headers.set("Content-Type"sv, mime_type);
  140. return response_headers;
  141. }
  142. void ResourceLoader::load(LoadRequest& request, SuccessCallback success_callback, ErrorCallback error_callback, Optional<u32> timeout, TimeoutCallback timeout_callback)
  143. {
  144. auto& url = request.url();
  145. request.start_timer();
  146. auto id = resource_id++;
  147. auto url_for_logging = sanitized_url_for_logging(url);
  148. emit_signpost(ByteString::formatted("Starting load: {}", url_for_logging), id);
  149. dbgln_if(SPAM_DEBUG, "ResourceLoader: Starting load of: \"{}\"", url_for_logging);
  150. auto const log_success = [url_for_logging, id](auto const& request) {
  151. auto load_time_ms = request.load_time().to_milliseconds();
  152. emit_signpost(ByteString::formatted("Finished load: {}", url_for_logging), id);
  153. dbgln_if(SPAM_DEBUG, "ResourceLoader: Finished load of: \"{}\", Duration: {}ms", url_for_logging, load_time_ms);
  154. };
  155. auto const log_failure = [url_for_logging, id](auto const& request, auto const& error_message) {
  156. auto load_time_ms = request.load_time().to_milliseconds();
  157. emit_signpost(ByteString::formatted("Failed load: {}", url_for_logging), id);
  158. dbgln("ResourceLoader: Failed load of: \"{}\", \033[31;1mError: {}\033[0m, Duration: {}ms", url_for_logging, error_message, load_time_ms);
  159. };
  160. auto respond_directory_page = [log_success, log_failure](LoadRequest const& request, URL::URL const& url, SuccessCallback const& success_callback, ErrorCallback const& error_callback) {
  161. auto maybe_response = load_file_directory_page(url);
  162. if (maybe_response.is_error()) {
  163. log_failure(request, maybe_response.error());
  164. if (error_callback)
  165. error_callback(ByteString::formatted("{}", maybe_response.error()), 500u, {}, {});
  166. return;
  167. }
  168. log_success(request);
  169. HashMap<ByteString, ByteString, CaseInsensitiveStringTraits> response_headers;
  170. response_headers.set("Content-Type"sv, "text/html"sv);
  171. success_callback(maybe_response.release_value().bytes(), response_headers, {});
  172. };
  173. if (is_port_blocked(url.port_or_default())) {
  174. log_failure(request, ByteString::formatted("The port #{} is blocked", url.port_or_default()));
  175. return;
  176. }
  177. if (ContentFilter::the().is_filtered(url)) {
  178. auto filter_message = "URL was filtered"sv;
  179. log_failure(request, filter_message);
  180. error_callback(filter_message, {}, {}, {});
  181. return;
  182. }
  183. if (url.scheme() == "about") {
  184. dbgln_if(SPAM_DEBUG, "Loading about: URL {}", url);
  185. log_success(request);
  186. HashMap<ByteString, ByteString, CaseInsensitiveStringTraits> response_headers;
  187. response_headers.set("Content-Type", "text/html; charset=UTF-8");
  188. // About version page
  189. if (url.path_segment_at_index(0) == "version") {
  190. success_callback(MUST(load_about_version_page()).bytes(), response_headers, {});
  191. return;
  192. }
  193. // Other about static HTML pages
  194. auto resource = Core::Resource::load_from_uri(MUST(String::formatted("resource://ladybird/{}.html", url.path_segment_at_index(0))));
  195. if (!resource.is_error()) {
  196. auto data = resource.value()->data();
  197. success_callback(data, response_headers, {});
  198. return;
  199. }
  200. Platform::EventLoopPlugin::the().deferred_invoke([success_callback = move(success_callback), response_headers = move(response_headers)] {
  201. success_callback(ByteString::empty().to_byte_buffer(), response_headers, {});
  202. });
  203. return;
  204. }
  205. if (url.scheme() == "data") {
  206. auto data_url_or_error = Fetch::Infrastructure::process_data_url(url);
  207. if (data_url_or_error.is_error()) {
  208. auto error_message = data_url_or_error.error().string_literal();
  209. log_failure(request, error_message);
  210. error_callback(error_message, {}, {}, {});
  211. return;
  212. }
  213. auto data_url = data_url_or_error.release_value();
  214. dbgln_if(SPAM_DEBUG, "ResourceLoader loading a data URL with mime-type: '{}', payload='{}'",
  215. data_url.mime_type,
  216. StringView(data_url.body.bytes()));
  217. HashMap<ByteString, ByteString, CaseInsensitiveStringTraits> response_headers;
  218. response_headers.set("Content-Type", data_url.mime_type.to_byte_string());
  219. log_success(request);
  220. Platform::EventLoopPlugin::the().deferred_invoke([data = move(data_url.body), response_headers = move(response_headers), success_callback = move(success_callback)] {
  221. success_callback(data, response_headers, {});
  222. });
  223. return;
  224. }
  225. if (url.scheme() == "resource") {
  226. auto resource = Core::Resource::load_from_uri(url.serialize());
  227. if (resource.is_error()) {
  228. log_failure(request, resource.error());
  229. return;
  230. }
  231. // When resource URI is a directory use file directory loader to generate response
  232. if (resource.value()->is_directory()) {
  233. respond_directory_page(request, resource.value()->file_url(), success_callback, error_callback);
  234. return;
  235. }
  236. auto data = resource.value()->data();
  237. auto response_headers = response_headers_for_file(url.serialize_path());
  238. log_success(request);
  239. success_callback(data, response_headers, {});
  240. return;
  241. }
  242. if (url.scheme() == "file") {
  243. if (request.page())
  244. m_page = request.page();
  245. if (!m_page.has_value()) {
  246. log_failure(request, "INTERNAL ERROR: No Page for request");
  247. return;
  248. }
  249. FileRequest file_request(url.serialize_path(), [this, success_callback = move(success_callback), error_callback = move(error_callback), log_success, log_failure, request, respond_directory_page](ErrorOr<i32> file_or_error) {
  250. --m_pending_loads;
  251. if (on_load_counter_change)
  252. on_load_counter_change();
  253. if (file_or_error.is_error()) {
  254. log_failure(request, file_or_error.error());
  255. if (error_callback) {
  256. auto status = file_or_error.error().code() == ENOENT ? 404u : 500u;
  257. error_callback(ByteString::formatted("{}", file_or_error.error()), status, {}, {});
  258. }
  259. return;
  260. }
  261. auto const fd = file_or_error.value();
  262. // When local file is a directory use file directory loader to generate response
  263. auto maybe_is_valid_directory = Core::Directory::is_valid_directory(fd);
  264. if (!maybe_is_valid_directory.is_error() && maybe_is_valid_directory.value()) {
  265. respond_directory_page(request, request.url(), success_callback, error_callback);
  266. return;
  267. }
  268. // Try to read file normally
  269. auto maybe_file = Core::File::adopt_fd(fd, Core::File::OpenMode::Read);
  270. if (maybe_file.is_error()) {
  271. log_failure(request, maybe_file.error());
  272. if (error_callback)
  273. error_callback(ByteString::formatted("{}", maybe_file.error()), 500u, {}, {});
  274. return;
  275. }
  276. auto file = maybe_file.release_value();
  277. auto maybe_data = file->read_until_eof();
  278. if (maybe_data.is_error()) {
  279. log_failure(request, maybe_data.error());
  280. if (error_callback)
  281. error_callback(ByteString::formatted("{}", maybe_data.error()), 500u, {}, {});
  282. return;
  283. }
  284. auto data = maybe_data.release_value();
  285. auto response_headers = response_headers_for_file(request.url().serialize_path());
  286. log_success(request);
  287. success_callback(data, response_headers, {});
  288. });
  289. (*m_page)->client().request_file(move(file_request));
  290. ++m_pending_loads;
  291. if (on_load_counter_change)
  292. on_load_counter_change();
  293. return;
  294. }
  295. if (url.scheme() == "http" || url.scheme() == "https" || url.scheme() == "gemini") {
  296. auto proxy = ProxyMappings::the().proxy_for_url(url);
  297. HashMap<ByteString, ByteString> headers;
  298. headers.set("User-Agent", m_user_agent.to_byte_string());
  299. headers.set("Accept-Encoding", "gzip, deflate, br");
  300. for (auto& it : request.headers()) {
  301. headers.set(it.key, it.value);
  302. }
  303. auto protocol_request = m_connector->start_request(request.method(), url, headers, request.body(), proxy);
  304. if (!protocol_request) {
  305. auto start_request_failure_msg = "Failed to initiate load"sv;
  306. log_failure(request, start_request_failure_msg);
  307. if (error_callback)
  308. error_callback(start_request_failure_msg, {}, {}, {});
  309. return;
  310. }
  311. if (timeout.has_value() && timeout.value() > 0) {
  312. auto timer = Platform::Timer::create_single_shot(timeout.value(), nullptr);
  313. timer->on_timeout = [timer, protocol_request, timeout_callback = move(timeout_callback)] {
  314. protocol_request->stop();
  315. if (timeout_callback)
  316. timeout_callback();
  317. };
  318. timer->start();
  319. }
  320. m_active_requests.set(*protocol_request);
  321. 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) mutable {
  322. --m_pending_loads;
  323. if (on_load_counter_change)
  324. on_load_counter_change();
  325. if (request.page()) {
  326. if (auto set_cookie = response_headers.get("Set-Cookie"); set_cookie.has_value())
  327. store_response_cookies(*request.page(), request.url(), *set_cookie);
  328. if (auto cache_control = response_headers.get("cache-control"); cache_control.has_value()) {
  329. if (cache_control.value().contains("no-store"sv)) {
  330. s_resource_cache.remove(request);
  331. }
  332. }
  333. }
  334. if (!success || (status_code.has_value() && *status_code >= 400 && *status_code <= 599 && (payload.is_empty() || !request.is_main_resource()))) {
  335. StringBuilder error_builder;
  336. if (status_code.has_value())
  337. error_builder.appendff("Load failed: {}", *status_code);
  338. else
  339. error_builder.append("Load failed"sv);
  340. log_failure(request, error_builder.string_view());
  341. if (error_callback)
  342. error_callback(error_builder.to_byte_string(), status_code, payload, response_headers);
  343. return;
  344. }
  345. log_success(request);
  346. success_callback(payload, response_headers, status_code);
  347. Platform::EventLoopPlugin::the().deferred_invoke([this, &protocol_request] {
  348. m_active_requests.remove(protocol_request);
  349. });
  350. };
  351. protocol_request->set_should_buffer_all_input(true);
  352. protocol_request->on_certificate_requested = []() -> ResourceLoaderConnectorRequest::CertificateAndKey {
  353. return {};
  354. };
  355. ++m_pending_loads;
  356. if (on_load_counter_change)
  357. on_load_counter_change();
  358. return;
  359. }
  360. auto not_implemented_error = ByteString::formatted("Protocol not implemented: {}", url.scheme());
  361. log_failure(request, not_implemented_error);
  362. if (error_callback)
  363. error_callback(not_implemented_error, {}, {}, {});
  364. }
  365. bool ResourceLoader::is_port_blocked(int port)
  366. {
  367. int ports[] { 1, 7, 9, 11, 13, 15, 17, 19, 20, 21, 22, 23, 25, 37, 42,
  368. 43, 53, 77, 79, 87, 95, 101, 102, 103, 104, 109, 110, 111, 113,
  369. 115, 117, 119, 123, 135, 139, 143, 179, 389, 465, 512, 513, 514,
  370. 515, 526, 530, 531, 532, 540, 556, 563, 587, 601, 636, 993, 995,
  371. 2049, 3659, 4045, 6000, 6379, 6665, 6666, 6667, 6668, 6669 };
  372. for (auto blocked_port : ports)
  373. if (port == blocked_port)
  374. return true;
  375. return false;
  376. }
  377. void ResourceLoader::clear_cache()
  378. {
  379. dbgln_if(CACHE_DEBUG, "Clearing {} items from ResourceLoader cache", s_resource_cache.size());
  380. s_resource_cache.clear();
  381. }
  382. void ResourceLoader::evict_from_cache(LoadRequest const& request)
  383. {
  384. dbgln_if(CACHE_DEBUG, "Removing resource {} from cache", request.url());
  385. s_resource_cache.remove(request);
  386. }
  387. }