ResourceLoader.cpp 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286
  1. /*
  2. * Copyright (c) 2018-2021, Andreas Kling <kling@serenityos.org>
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. #include <AK/Base64.h>
  7. #include <AK/Debug.h>
  8. #include <AK/JsonObject.h>
  9. #include <LibCore/ElapsedTimer.h>
  10. #include <LibCore/EventLoop.h>
  11. #include <LibCore/File.h>
  12. #include <LibProtocol/Request.h>
  13. #include <LibProtocol/RequestClient.h>
  14. #include <LibWeb/Loader/ContentFilter.h>
  15. #include <LibWeb/Loader/LoadRequest.h>
  16. #include <LibWeb/Loader/ProxyMappings.h>
  17. #include <LibWeb/Loader/Resource.h>
  18. #include <LibWeb/Loader/ResourceLoader.h>
  19. #include <serenity.h>
  20. namespace Web {
  21. ResourceLoader& ResourceLoader::the()
  22. {
  23. static RefPtr<ResourceLoader> s_the;
  24. if (!s_the)
  25. s_the = ResourceLoader::try_create().release_value_but_fixme_should_propagate_errors();
  26. return *s_the;
  27. }
  28. ErrorOr<NonnullRefPtr<ResourceLoader>> ResourceLoader::try_create()
  29. {
  30. auto protocol_client = TRY(Protocol::RequestClient::try_create());
  31. return adopt_nonnull_ref_or_enomem(new (nothrow) ResourceLoader(move(protocol_client)));
  32. }
  33. ResourceLoader::ResourceLoader(NonnullRefPtr<Protocol::RequestClient> protocol_client)
  34. : m_protocol_client(move(protocol_client))
  35. , m_user_agent(default_user_agent)
  36. {
  37. }
  38. void ResourceLoader::prefetch_dns(AK::URL const& url)
  39. {
  40. m_protocol_client->ensure_connection(url, RequestServer::CacheLevel::ResolveOnly);
  41. }
  42. void ResourceLoader::preconnect(AK::URL const& url)
  43. {
  44. m_protocol_client->ensure_connection(url, RequestServer::CacheLevel::CreateConnection);
  45. }
  46. static HashMap<LoadRequest, NonnullRefPtr<Resource>> s_resource_cache;
  47. RefPtr<Resource> ResourceLoader::load_resource(Resource::Type type, LoadRequest& request)
  48. {
  49. if (!request.is_valid())
  50. return nullptr;
  51. bool use_cache = request.url().protocol() != "file";
  52. if (use_cache) {
  53. auto it = s_resource_cache.find(request);
  54. if (it != s_resource_cache.end()) {
  55. if (it->value->type() != type) {
  56. dbgln("FIXME: Not using cached resource for {} since there's a type mismatch.", request.url());
  57. } else {
  58. dbgln_if(CACHE_DEBUG, "Reusing cached resource for: {}", request.url());
  59. return it->value;
  60. }
  61. }
  62. }
  63. auto resource = Resource::create({}, type, request);
  64. if (use_cache)
  65. s_resource_cache.set(request, resource);
  66. load(
  67. request,
  68. [=](auto data, auto& headers, auto status_code) {
  69. const_cast<Resource&>(*resource).did_load({}, data, headers, status_code);
  70. },
  71. [=](auto& error, auto status_code) {
  72. const_cast<Resource&>(*resource).did_fail({}, error, status_code);
  73. });
  74. return resource;
  75. }
  76. static String sanitized_url_for_logging(AK::URL const& url)
  77. {
  78. if (url.protocol() == "data"sv)
  79. return String::formatted("[data URL, mime-type={}, size={}]", url.data_mime_type(), url.data_payload().length());
  80. return url.to_string();
  81. }
  82. static void emit_signpost(String const& message, int id)
  83. {
  84. auto string_id = perf_register_string(message.characters(), message.length());
  85. perf_event(PERF_EVENT_SIGNPOST, string_id, id);
  86. }
  87. static size_t resource_id = 0;
  88. 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)
  89. {
  90. auto& url = request.url();
  91. request.start_timer();
  92. auto id = resource_id++;
  93. auto url_for_logging = sanitized_url_for_logging(url);
  94. emit_signpost(String::formatted("Starting load: {}", url_for_logging), id);
  95. dbgln("ResourceLoader: Starting load of: \"{}\"", url_for_logging);
  96. auto const log_success = [url_for_logging, id](auto const& request) {
  97. auto load_time_ms = request.load_time().to_milliseconds();
  98. emit_signpost(String::formatted("Finished load: {}", url_for_logging), id);
  99. dbgln("ResourceLoader: Finished load of: \"{}\", Duration: {}ms", url_for_logging, load_time_ms);
  100. };
  101. auto const log_failure = [url_for_logging, id](auto const& request, auto const error_message) {
  102. auto load_time_ms = request.load_time().to_milliseconds();
  103. emit_signpost(String::formatted("Failed load: {}", url_for_logging), id);
  104. dbgln("ResourceLoader: Failed load of: \"{}\", \033[31;1mError: {}\033[0m, Duration: {}ms", url_for_logging, error_message, load_time_ms);
  105. };
  106. if (is_port_blocked(url.port_or_default())) {
  107. log_failure(request, String::formatted("The port #{} is blocked", url.port_or_default()));
  108. return;
  109. }
  110. if (ContentFilter::the().is_filtered(url)) {
  111. auto filter_message = "URL was filtered"sv;
  112. log_failure(request, filter_message);
  113. error_callback(filter_message, {});
  114. return;
  115. }
  116. if (url.protocol() == "about") {
  117. dbgln_if(SPAM_DEBUG, "Loading about: URL {}", url);
  118. log_success(request);
  119. deferred_invoke([success_callback = move(success_callback)] {
  120. success_callback(String::empty().to_byte_buffer(), {}, {});
  121. });
  122. return;
  123. }
  124. if (url.protocol() == "data") {
  125. dbgln_if(SPAM_DEBUG, "ResourceLoader loading a data URL with mime-type: '{}', base64={}, payload='{}'",
  126. url.data_mime_type(),
  127. url.data_payload_is_base64(),
  128. url.data_payload());
  129. ByteBuffer data;
  130. if (url.data_payload_is_base64()) {
  131. auto data_maybe = decode_base64(url.data_payload());
  132. if (data_maybe.is_error()) {
  133. auto error_message = data_maybe.error().string_literal();
  134. log_failure(request, error_message);
  135. error_callback(error_message, {});
  136. return;
  137. }
  138. data = data_maybe.value();
  139. } else {
  140. data = url.data_payload().to_byte_buffer();
  141. }
  142. log_success(request);
  143. deferred_invoke([data = move(data), success_callback = move(success_callback)] {
  144. success_callback(data, {}, {});
  145. });
  146. return;
  147. }
  148. if (url.protocol() == "file") {
  149. auto file_result = Core::File::open(url.path(), Core::OpenMode::ReadOnly);
  150. if (file_result.is_error()) {
  151. auto& error = file_result.error();
  152. log_failure(request, error);
  153. if (error_callback)
  154. error_callback(String::formatted("{}", error), error.code());
  155. return;
  156. }
  157. auto file = file_result.release_value();
  158. auto data = file->read_all();
  159. log_success(request);
  160. deferred_invoke([data = move(data), success_callback = move(success_callback)] {
  161. success_callback(data, {}, {});
  162. });
  163. return;
  164. }
  165. if (url.protocol() == "http" || url.protocol() == "https" || url.protocol() == "gemini") {
  166. auto proxy = ProxyMappings::the().proxy_for_url(url);
  167. HashMap<String, String> headers;
  168. headers.set("User-Agent", m_user_agent);
  169. headers.set("Accept-Encoding", "gzip, deflate");
  170. for (auto& it : request.headers()) {
  171. headers.set(it.key, it.value);
  172. }
  173. auto protocol_request = protocol_client().start_request(request.method(), url, headers, request.body(), proxy);
  174. if (!protocol_request) {
  175. auto start_request_failure_msg = "Failed to initiate load"sv;
  176. log_failure(request, start_request_failure_msg);
  177. if (error_callback)
  178. error_callback(start_request_failure_msg, {});
  179. return;
  180. }
  181. m_active_requests.set(*protocol_request);
  182. 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) {
  183. --m_pending_loads;
  184. if (on_load_counter_change)
  185. on_load_counter_change();
  186. if (!success || (status_code.has_value() && *status_code >= 400 && *status_code <= 599)) {
  187. StringBuilder error_builder;
  188. if (status_code.has_value())
  189. error_builder.appendff("Load failed: {}", *status_code);
  190. else
  191. error_builder.append("Load failed");
  192. log_failure(request, error_builder.string_view());
  193. if (error_callback)
  194. error_callback(error_builder.to_string(), {});
  195. return;
  196. }
  197. log_success(request);
  198. success_callback(payload, response_headers, status_code);
  199. deferred_invoke([this, &protocol_request] {
  200. m_active_requests.remove(protocol_request);
  201. });
  202. };
  203. protocol_request->set_should_buffer_all_input(true);
  204. protocol_request->on_certificate_requested = []() -> Protocol::Request::CertificateAndKey {
  205. return {};
  206. };
  207. ++m_pending_loads;
  208. if (on_load_counter_change)
  209. on_load_counter_change();
  210. return;
  211. }
  212. auto not_implemented_error = String::formatted("Protocol not implemented: {}", url.protocol());
  213. log_failure(request, not_implemented_error);
  214. if (error_callback)
  215. error_callback(not_implemented_error, {});
  216. }
  217. 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)
  218. {
  219. LoadRequest request;
  220. request.set_url(url);
  221. load(request, move(success_callback), move(error_callback));
  222. }
  223. bool ResourceLoader::is_port_blocked(int port)
  224. {
  225. int ports[] { 1, 7, 9, 11, 13, 15, 17, 19, 20, 21, 22, 23, 25, 37, 42,
  226. 43, 53, 77, 79, 87, 95, 101, 102, 103, 104, 109, 110, 111, 113,
  227. 115, 117, 119, 123, 135, 139, 143, 179, 389, 465, 512, 513, 514,
  228. 515, 526, 530, 531, 532, 540, 556, 563, 587, 601, 636, 993, 995,
  229. 2049, 3659, 4045, 6000, 6379, 6665, 6666, 6667, 6668, 6669, 9000 };
  230. for (auto blocked_port : ports)
  231. if (port == blocked_port)
  232. return true;
  233. return false;
  234. }
  235. void ResourceLoader::clear_cache()
  236. {
  237. dbgln_if(CACHE_DEBUG, "Clearing {} items from ResourceLoader cache", s_resource_cache.size());
  238. s_resource_cache.clear();
  239. }
  240. void ResourceLoader::evict_from_cache(LoadRequest const& request)
  241. {
  242. dbgln_if(CACHE_DEBUG, "Removing resource {} from cache", request.url());
  243. s_resource_cache.remove(request);
  244. }
  245. }