ResourceLoader.cpp 18 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471
  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/Loader/ContentFilter.h>
  16. #include <LibWeb/Loader/GeneratedPagesLoader.h>
  17. #include <LibWeb/Loader/LoadRequest.h>
  18. #include <LibWeb/Loader/ProxyMappings.h>
  19. #include <LibWeb/Loader/Resource.h>
  20. #include <LibWeb/Loader/ResourceLoader.h>
  21. #include <LibWeb/Platform/EventLoopPlugin.h>
  22. #include <LibWeb/Platform/Timer.h>
  23. #ifdef AK_OS_SERENITY
  24. # include <serenity.h>
  25. #endif
  26. namespace Web {
  27. ResourceLoaderConnectorRequest::ResourceLoaderConnectorRequest() = default;
  28. ResourceLoaderConnectorRequest::~ResourceLoaderConnectorRequest() = default;
  29. ResourceLoaderConnector::ResourceLoaderConnector() = default;
  30. ResourceLoaderConnector::~ResourceLoaderConnector() = default;
  31. static RefPtr<ResourceLoader> s_resource_loader;
  32. void ResourceLoader::initialize(RefPtr<ResourceLoaderConnector> connector)
  33. {
  34. if (connector)
  35. s_resource_loader = ResourceLoader::try_create(connector.release_nonnull()).release_value_but_fixme_should_propagate_errors();
  36. }
  37. ResourceLoader& ResourceLoader::the()
  38. {
  39. if (!s_resource_loader) {
  40. dbgln("Web::ResourceLoader was not initialized");
  41. VERIFY_NOT_REACHED();
  42. }
  43. return *s_resource_loader;
  44. }
  45. ErrorOr<NonnullRefPtr<ResourceLoader>> ResourceLoader::try_create(NonnullRefPtr<ResourceLoaderConnector> connector)
  46. {
  47. return adopt_nonnull_ref_or_enomem(new (nothrow) ResourceLoader(move(connector)));
  48. }
  49. ResourceLoader::ResourceLoader(NonnullRefPtr<ResourceLoaderConnector> connector)
  50. : m_connector(move(connector))
  51. , m_user_agent(MUST(String::from_utf8(default_user_agent)))
  52. , m_platform(MUST(String::from_utf8(default_platform)))
  53. {
  54. }
  55. void ResourceLoader::prefetch_dns(URL const& url)
  56. {
  57. if (ContentFilter::the().is_filtered(url)) {
  58. dbgln("ResourceLoader: Refusing to prefetch DNS for '{}': \033[31;1mURL was filtered\033[0m", url);
  59. return;
  60. }
  61. m_connector->prefetch_dns(url);
  62. }
  63. void ResourceLoader::preconnect(URL const& url)
  64. {
  65. if (url.scheme().is_one_of("file"sv, "data"sv))
  66. return;
  67. if (ContentFilter::the().is_filtered(url)) {
  68. dbgln("ResourceLoader: Refusing to pre-connect to '{}': \033[31;1mURL was filtered\033[0m", url);
  69. return;
  70. }
  71. m_connector->preconnect(url);
  72. }
  73. static HashMap<LoadRequest, NonnullRefPtr<Resource>> s_resource_cache;
  74. RefPtr<Resource> ResourceLoader::load_resource(Resource::Type type, LoadRequest& request)
  75. {
  76. if (!request.is_valid())
  77. return nullptr;
  78. bool use_cache = request.url().scheme() != "file";
  79. if (use_cache) {
  80. auto it = s_resource_cache.find(request);
  81. if (it != s_resource_cache.end()) {
  82. if (it->value->type() != type) {
  83. dbgln("FIXME: Not using cached resource for {} since there's a type mismatch.", request.url());
  84. } else {
  85. dbgln_if(CACHE_DEBUG, "Reusing cached resource for: {}", request.url());
  86. return it->value;
  87. }
  88. }
  89. }
  90. auto resource = Resource::create({}, type, request);
  91. if (use_cache)
  92. s_resource_cache.set(request, resource);
  93. load(
  94. request,
  95. [=](auto data, auto& headers, auto status_code) {
  96. const_cast<Resource&>(*resource).did_load({}, data, headers, status_code);
  97. },
  98. [=](auto& error, auto status_code, auto, auto) {
  99. const_cast<Resource&>(*resource).did_fail({}, error, status_code);
  100. });
  101. return resource;
  102. }
  103. static ByteString sanitized_url_for_logging(URL const& url)
  104. {
  105. if (url.scheme() == "data"sv)
  106. return "[data URL]"sv;
  107. return url.to_byte_string();
  108. }
  109. static void emit_signpost(ByteString const& message, int id)
  110. {
  111. #ifdef AK_OS_SERENITY
  112. auto string_id = perf_register_string(message.characters(), message.length());
  113. perf_event(PERF_EVENT_SIGNPOST, string_id, id);
  114. #else
  115. (void)message;
  116. (void)id;
  117. #endif
  118. }
  119. static void store_response_cookies(Page& page, URL const& url, ByteString const& cookies)
  120. {
  121. auto set_cookie_json_value = MUST(JsonValue::from_string(cookies));
  122. VERIFY(set_cookie_json_value.type() == JsonValue::Type::Array);
  123. for (auto const& set_cookie_entry : set_cookie_json_value.as_array().values()) {
  124. VERIFY(set_cookie_entry.type() == JsonValue::Type::String);
  125. auto cookie = Cookie::parse_cookie(set_cookie_entry.as_string());
  126. if (!cookie.has_value())
  127. continue;
  128. page.client().page_did_set_cookie(url, cookie.value(), Cookie::Source::Http); // FIXME: Determine cookie source correctly
  129. }
  130. }
  131. static size_t resource_id = 0;
  132. static HashMap<ByteString, ByteString, CaseInsensitiveStringTraits> response_headers_for_file(StringView path)
  133. {
  134. // For file:// and resource:// URLs, we have to guess the MIME type, since there's no HTTP header to tell us what
  135. // it is. We insert a fake Content-Type header here, so that clients can use it to learn the MIME type.
  136. auto mime_type = Core::guess_mime_type_based_on_filename(path);
  137. HashMap<ByteString, ByteString, CaseInsensitiveStringTraits> response_headers;
  138. response_headers.set("Content-Type"sv, mime_type);
  139. return response_headers;
  140. }
  141. void ResourceLoader::load(LoadRequest& request, SuccessCallback success_callback, ErrorCallback error_callback, Optional<u32> timeout, TimeoutCallback timeout_callback)
  142. {
  143. auto& url = request.url();
  144. request.start_timer();
  145. auto id = resource_id++;
  146. auto url_for_logging = sanitized_url_for_logging(url);
  147. emit_signpost(ByteString::formatted("Starting load: {}", url_for_logging), id);
  148. dbgln_if(SPAM_DEBUG, "ResourceLoader: Starting load of: \"{}\"", url_for_logging);
  149. auto const log_success = [url_for_logging, id](auto const& request) {
  150. auto load_time_ms = request.load_time().to_milliseconds();
  151. emit_signpost(ByteString::formatted("Finished load: {}", url_for_logging), id);
  152. dbgln_if(SPAM_DEBUG, "ResourceLoader: Finished load of: \"{}\", Duration: {}ms", url_for_logging, load_time_ms);
  153. };
  154. auto const log_failure = [url_for_logging, id](auto const& request, auto const& error_message) {
  155. auto load_time_ms = request.load_time().to_milliseconds();
  156. emit_signpost(ByteString::formatted("Failed load: {}", url_for_logging), id);
  157. dbgln("ResourceLoader: Failed load of: \"{}\", \033[31;1mError: {}\033[0m, Duration: {}ms", url_for_logging, error_message, load_time_ms);
  158. };
  159. auto respond_directory_page = [log_success, log_failure](LoadRequest const& request, URL const& url, SuccessCallback const& success_callback, ErrorCallback const& error_callback) {
  160. auto maybe_response = load_file_directory_page(url);
  161. if (maybe_response.is_error()) {
  162. log_failure(request, maybe_response.error());
  163. if (error_callback)
  164. error_callback(ByteString::formatted("{}", maybe_response.error()), 500u, {}, {});
  165. return;
  166. }
  167. log_success(request);
  168. HashMap<ByteString, ByteString, CaseInsensitiveStringTraits> response_headers;
  169. response_headers.set("Content-Type"sv, "text/html"sv);
  170. success_callback(maybe_response.release_value().bytes(), response_headers, {});
  171. };
  172. if (is_port_blocked(url.port_or_default())) {
  173. log_failure(request, ByteString::formatted("The port #{} is blocked", url.port_or_default()));
  174. return;
  175. }
  176. if (ContentFilter::the().is_filtered(url)) {
  177. auto filter_message = "URL was filtered"sv;
  178. log_failure(request, filter_message);
  179. error_callback(filter_message, {}, {}, {});
  180. return;
  181. }
  182. if (url.scheme() == "about") {
  183. dbgln_if(SPAM_DEBUG, "Loading about: URL {}", url);
  184. log_success(request);
  185. HashMap<ByteString, ByteString, CaseInsensitiveStringTraits> response_headers;
  186. response_headers.set("Content-Type", "text/html; charset=UTF-8");
  187. // About version page
  188. if (url.path_segment_at_index(0) == "version") {
  189. success_callback(MUST(load_about_version_page()).bytes(), response_headers, {});
  190. return;
  191. }
  192. // Other about static HTML pages
  193. auto resource = Core::Resource::load_from_uri(MUST(String::formatted("resource://ladybird/{}.html", url.path_segment_at_index(0))));
  194. if (!resource.is_error()) {
  195. auto data = resource.value()->data();
  196. success_callback(data, response_headers, {});
  197. return;
  198. }
  199. Platform::EventLoopPlugin::the().deferred_invoke([success_callback = move(success_callback), response_headers = move(response_headers)] {
  200. success_callback(ByteString::empty().to_byte_buffer(), response_headers, {});
  201. });
  202. return;
  203. }
  204. if (url.scheme() == "data") {
  205. auto data_url_or_error = url.process_data_url();
  206. if (data_url_or_error.is_error()) {
  207. auto error_message = data_url_or_error.error().string_literal();
  208. log_failure(request, error_message);
  209. error_callback(error_message, {}, {}, {});
  210. return;
  211. }
  212. auto data_url = data_url_or_error.release_value();
  213. dbgln_if(SPAM_DEBUG, "ResourceLoader loading a data URL with mime-type: '{}', payload='{}'",
  214. data_url.mime_type,
  215. StringView(data_url.body.bytes()));
  216. HashMap<ByteString, ByteString, CaseInsensitiveStringTraits> response_headers;
  217. response_headers.set("Content-Type", data_url.mime_type.to_byte_string());
  218. log_success(request);
  219. Platform::EventLoopPlugin::the().deferred_invoke([data = move(data_url.body), response_headers = move(response_headers), success_callback = move(success_callback)] {
  220. success_callback(data, response_headers, {});
  221. });
  222. return;
  223. }
  224. if (url.scheme() == "resource") {
  225. auto resource = Core::Resource::load_from_uri(url.serialize());
  226. if (resource.is_error()) {
  227. log_failure(request, resource.error());
  228. return;
  229. }
  230. // When resource URI is a directory use file directory loader to generate response
  231. if (resource.value()->is_directory()) {
  232. respond_directory_page(request, resource.value()->file_url(), success_callback, error_callback);
  233. return;
  234. }
  235. auto data = resource.value()->data();
  236. auto response_headers = response_headers_for_file(url.serialize_path());
  237. log_success(request);
  238. success_callback(data, response_headers, {});
  239. return;
  240. }
  241. if (url.scheme() == "file") {
  242. if (request.page())
  243. m_page = request.page();
  244. if (!m_page.has_value()) {
  245. log_failure(request, "INTERNAL ERROR: No Page for request");
  246. return;
  247. }
  248. 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) {
  249. --m_pending_loads;
  250. if (on_load_counter_change)
  251. on_load_counter_change();
  252. if (file_or_error.is_error()) {
  253. log_failure(request, file_or_error.error());
  254. if (error_callback) {
  255. auto status = file_or_error.error().code() == ENOENT ? 404u : 500u;
  256. error_callback(ByteString::formatted("{}", file_or_error.error()), status, {}, {});
  257. }
  258. return;
  259. }
  260. auto const fd = file_or_error.value();
  261. // When local file is a directory use file directory loader to generate response
  262. auto maybe_is_valid_directory = Core::Directory::is_valid_directory(fd);
  263. if (!maybe_is_valid_directory.is_error() && maybe_is_valid_directory.value()) {
  264. respond_directory_page(request, request.url(), success_callback, error_callback);
  265. return;
  266. }
  267. // Try to read file normally
  268. auto maybe_file = Core::File::adopt_fd(fd, Core::File::OpenMode::Read);
  269. if (maybe_file.is_error()) {
  270. log_failure(request, maybe_file.error());
  271. if (error_callback)
  272. error_callback(ByteString::formatted("{}", maybe_file.error()), 500u, {}, {});
  273. return;
  274. }
  275. auto file = maybe_file.release_value();
  276. auto maybe_data = file->read_until_eof();
  277. if (maybe_data.is_error()) {
  278. log_failure(request, maybe_data.error());
  279. if (error_callback)
  280. error_callback(ByteString::formatted("{}", maybe_data.error()), 500u, {}, {});
  281. return;
  282. }
  283. auto data = maybe_data.release_value();
  284. auto response_headers = response_headers_for_file(request.url().serialize_path());
  285. log_success(request);
  286. success_callback(data, response_headers, {});
  287. });
  288. (*m_page)->client().request_file(move(file_request));
  289. ++m_pending_loads;
  290. if (on_load_counter_change)
  291. on_load_counter_change();
  292. return;
  293. }
  294. if (url.scheme() == "http" || url.scheme() == "https" || url.scheme() == "gemini") {
  295. auto proxy = ProxyMappings::the().proxy_for_url(url);
  296. HashMap<ByteString, ByteString> headers;
  297. headers.set("User-Agent", m_user_agent.to_byte_string());
  298. headers.set("Accept-Encoding", "gzip, deflate, br");
  299. for (auto& it : request.headers()) {
  300. headers.set(it.key, it.value);
  301. }
  302. auto protocol_request = m_connector->start_request(request.method(), url, headers, request.body(), proxy);
  303. if (!protocol_request) {
  304. auto start_request_failure_msg = "Failed to initiate load"sv;
  305. log_failure(request, start_request_failure_msg);
  306. if (error_callback)
  307. error_callback(start_request_failure_msg, {}, {}, {});
  308. return;
  309. }
  310. if (timeout.has_value() && timeout.value() > 0) {
  311. auto timer = Platform::Timer::create_single_shot(timeout.value(), nullptr);
  312. timer->on_timeout = [timer, protocol_request, timeout_callback = move(timeout_callback)] {
  313. protocol_request->stop();
  314. if (timeout_callback)
  315. timeout_callback();
  316. };
  317. timer->start();
  318. }
  319. m_active_requests.set(*protocol_request);
  320. 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 {
  321. --m_pending_loads;
  322. if (on_load_counter_change)
  323. on_load_counter_change();
  324. if (request.page()) {
  325. if (auto set_cookie = response_headers.get("Set-Cookie"); set_cookie.has_value())
  326. store_response_cookies(*request.page(), request.url(), *set_cookie);
  327. if (auto cache_control = response_headers.get("cache-control"); cache_control.has_value()) {
  328. if (cache_control.value().contains("no-store"sv)) {
  329. s_resource_cache.remove(request);
  330. }
  331. }
  332. }
  333. if (!success || (status_code.has_value() && *status_code >= 400 && *status_code <= 599 && (payload.is_empty() || !request.is_main_resource()))) {
  334. StringBuilder error_builder;
  335. if (status_code.has_value())
  336. error_builder.appendff("Load failed: {}", *status_code);
  337. else
  338. error_builder.append("Load failed"sv);
  339. log_failure(request, error_builder.string_view());
  340. if (error_callback)
  341. error_callback(error_builder.to_byte_string(), status_code, payload, response_headers);
  342. return;
  343. }
  344. log_success(request);
  345. success_callback(payload, response_headers, status_code);
  346. Platform::EventLoopPlugin::the().deferred_invoke([this, &protocol_request] {
  347. m_active_requests.remove(protocol_request);
  348. });
  349. };
  350. protocol_request->set_should_buffer_all_input(true);
  351. protocol_request->on_certificate_requested = []() -> ResourceLoaderConnectorRequest::CertificateAndKey {
  352. return {};
  353. };
  354. ++m_pending_loads;
  355. if (on_load_counter_change)
  356. on_load_counter_change();
  357. return;
  358. }
  359. auto not_implemented_error = ByteString::formatted("Protocol not implemented: {}", url.scheme());
  360. log_failure(request, not_implemented_error);
  361. if (error_callback)
  362. error_callback(not_implemented_error, {}, {}, {});
  363. }
  364. bool ResourceLoader::is_port_blocked(int port)
  365. {
  366. int ports[] { 1, 7, 9, 11, 13, 15, 17, 19, 20, 21, 22, 23, 25, 37, 42,
  367. 43, 53, 77, 79, 87, 95, 101, 102, 103, 104, 109, 110, 111, 113,
  368. 115, 117, 119, 123, 135, 139, 143, 179, 389, 465, 512, 513, 514,
  369. 515, 526, 530, 531, 532, 540, 556, 563, 587, 601, 636, 993, 995,
  370. 2049, 3659, 4045, 6000, 6379, 6665, 6666, 6667, 6668, 6669 };
  371. for (auto blocked_port : ports)
  372. if (port == blocked_port)
  373. return true;
  374. return false;
  375. }
  376. void ResourceLoader::clear_cache()
  377. {
  378. dbgln_if(CACHE_DEBUG, "Clearing {} items from ResourceLoader cache", s_resource_cache.size());
  379. s_resource_cache.clear();
  380. }
  381. void ResourceLoader::evict_from_cache(LoadRequest const& request)
  382. {
  383. dbgln_if(CACHE_DEBUG, "Removing resource {} from cache", request.url());
  384. s_resource_cache.remove(request);
  385. }
  386. }