ResourceLoader.cpp 10 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283
  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/Resource.h>
  17. #include <LibWeb/Loader/ResourceLoader.h>
  18. namespace Web {
  19. ResourceLoader& ResourceLoader::the()
  20. {
  21. static RefPtr<ResourceLoader> s_the;
  22. if (!s_the)
  23. s_the = ResourceLoader::try_create().release_value_but_fixme_should_propagate_errors();
  24. return *s_the;
  25. }
  26. ErrorOr<NonnullRefPtr<ResourceLoader>> ResourceLoader::try_create()
  27. {
  28. auto protocol_client = TRY(Protocol::RequestClient::try_create());
  29. return adopt_nonnull_ref_or_enomem(new (nothrow) ResourceLoader(move(protocol_client)));
  30. }
  31. ResourceLoader::ResourceLoader(NonnullRefPtr<Protocol::RequestClient> protocol_client)
  32. : m_protocol_client(move(protocol_client))
  33. , m_user_agent(default_user_agent)
  34. {
  35. }
  36. void ResourceLoader::load_sync(LoadRequest& request, Function<void(ReadonlyBytes, const HashMap<String, String, CaseInsensitiveStringTraits>& response_headers, Optional<u32> status_code)> success_callback, Function<void(const String&, Optional<u32> status_code)> error_callback)
  37. {
  38. Core::EventLoop loop;
  39. load(
  40. request,
  41. [&](auto data, auto& response_headers, auto status_code) {
  42. success_callback(data, response_headers, status_code);
  43. loop.quit(0);
  44. },
  45. [&](auto& string, auto status_code) {
  46. if (error_callback)
  47. error_callback(string, status_code);
  48. loop.quit(0);
  49. });
  50. loop.exec();
  51. }
  52. void ResourceLoader::prefetch_dns(AK::URL const& url)
  53. {
  54. m_protocol_client->ensure_connection(url, RequestServer::CacheLevel::ResolveOnly);
  55. }
  56. void ResourceLoader::preconnect(AK::URL const& url)
  57. {
  58. m_protocol_client->ensure_connection(url, RequestServer::CacheLevel::CreateConnection);
  59. }
  60. static HashMap<LoadRequest, NonnullRefPtr<Resource>> s_resource_cache;
  61. RefPtr<Resource> ResourceLoader::load_resource(Resource::Type type, LoadRequest& request)
  62. {
  63. if (!request.is_valid())
  64. return nullptr;
  65. bool use_cache = request.url().protocol() != "file";
  66. if (use_cache) {
  67. auto it = s_resource_cache.find(request);
  68. if (it != s_resource_cache.end()) {
  69. if (it->value->type() != type) {
  70. dbgln("FIXME: Not using cached resource for {} since there's a type mismatch.", request.url());
  71. } else {
  72. dbgln_if(CACHE_DEBUG, "Reusing cached resource for: {}", request.url());
  73. return it->value;
  74. }
  75. }
  76. }
  77. auto resource = Resource::create({}, type, request);
  78. if (use_cache)
  79. s_resource_cache.set(request, resource);
  80. load(
  81. request,
  82. [=](auto data, auto& headers, auto status_code) {
  83. const_cast<Resource&>(*resource).did_load({}, data, headers, status_code);
  84. },
  85. [=](auto& error, auto status_code) {
  86. const_cast<Resource&>(*resource).did_fail({}, error, status_code);
  87. });
  88. return resource;
  89. }
  90. static String sanitized_url_for_logging(AK::URL const& url)
  91. {
  92. if (url.protocol() == "data"sv)
  93. return String::formatted("[data URL, mime-type={}, size={}]", url.data_mime_type(), url.data_payload().length());
  94. return url.to_string();
  95. }
  96. void ResourceLoader::load(LoadRequest& request, Function<void(ReadonlyBytes, const HashMap<String, String, CaseInsensitiveStringTraits>& response_headers, Optional<u32> status_code)> success_callback, Function<void(const String&, Optional<u32> status_code)> error_callback)
  97. {
  98. auto& url = request.url();
  99. request.start_timer();
  100. dbgln("ResourceLoader: Starting load of: \"{}\"", sanitized_url_for_logging(url));
  101. const auto log_success = [](const auto& request) {
  102. auto& url = request.url();
  103. auto load_time_ms = request.load_time().to_milliseconds();
  104. dbgln("ResourceLoader: Finished load of: \"{}\", Duration: {}ms", sanitized_url_for_logging(url), load_time_ms);
  105. };
  106. const auto log_failure = [](const auto& request, const auto error_message) {
  107. auto& url = request.url();
  108. auto load_time_ms = request.load_time().to_milliseconds();
  109. dbgln("ResourceLoader: Failed load of: \"{}\", \033[31;1mError: {}\033[0m, Duration: {}ms", sanitized_url_for_logging(url), error_message, load_time_ms);
  110. };
  111. if (is_port_blocked(url.port_or_default())) {
  112. log_failure(request, String::formatted("The port #{} is blocked", url.port_or_default()));
  113. return;
  114. }
  115. if (ContentFilter::the().is_filtered(url)) {
  116. auto filter_message = "URL was filtered"sv;
  117. log_failure(request, filter_message);
  118. error_callback(filter_message, {});
  119. return;
  120. }
  121. if (url.protocol() == "about") {
  122. dbgln_if(SPAM_DEBUG, "Loading about: URL {}", url);
  123. log_success(request);
  124. deferred_invoke([success_callback = move(success_callback)] {
  125. success_callback(String::empty().to_byte_buffer(), {}, {});
  126. });
  127. return;
  128. }
  129. if (url.protocol() == "data") {
  130. dbgln_if(SPAM_DEBUG, "ResourceLoader loading a data URL with mime-type: '{}', base64={}, payload='{}'",
  131. url.data_mime_type(),
  132. url.data_payload_is_base64(),
  133. url.data_payload());
  134. ByteBuffer data;
  135. if (url.data_payload_is_base64()) {
  136. auto data_maybe = decode_base64(url.data_payload());
  137. if (data_maybe.is_error()) {
  138. auto error_message = data_maybe.error().string_literal();
  139. log_failure(request, error_message);
  140. error_callback(error_message, {});
  141. return;
  142. }
  143. data = data_maybe.value();
  144. } else {
  145. data = url.data_payload().to_byte_buffer();
  146. }
  147. log_success(request);
  148. deferred_invoke([data = move(data), success_callback = move(success_callback)] {
  149. success_callback(data, {}, {});
  150. });
  151. return;
  152. }
  153. if (url.protocol() == "file") {
  154. auto file_result = Core::File::open(url.path(), Core::OpenMode::ReadOnly);
  155. if (file_result.is_error()) {
  156. auto& error = file_result.error();
  157. log_failure(request, error);
  158. if (error_callback)
  159. error_callback(String::formatted("{}", error), error.code());
  160. return;
  161. }
  162. auto file = file_result.release_value();
  163. auto data = file->read_all();
  164. log_success(request);
  165. deferred_invoke([data = move(data), success_callback = move(success_callback)] {
  166. success_callback(data, {}, {});
  167. });
  168. return;
  169. }
  170. if (url.protocol() == "http" || url.protocol() == "https" || url.protocol() == "gemini") {
  171. HashMap<String, String> headers;
  172. headers.set("User-Agent", m_user_agent);
  173. headers.set("Accept-Encoding", "gzip, deflate");
  174. for (auto& it : request.headers()) {
  175. headers.set(it.key, it.value);
  176. }
  177. auto protocol_request = protocol_client().start_request(request.method(), url, headers, request.body());
  178. if (!protocol_request) {
  179. auto start_request_failure_msg = "Failed to initiate load"sv;
  180. log_failure(request, start_request_failure_msg);
  181. if (error_callback)
  182. error_callback(start_request_failure_msg, {});
  183. return;
  184. }
  185. m_active_requests.set(*protocol_request);
  186. 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) {
  187. --m_pending_loads;
  188. if (on_load_counter_change)
  189. on_load_counter_change();
  190. if (!success || (status_code.has_value() && *status_code >= 400 && *status_code <= 599)) {
  191. StringBuilder error_builder;
  192. if (status_code.has_value())
  193. error_builder.appendff("Load failed: {}", *status_code);
  194. else
  195. error_builder.append("Load failed");
  196. log_failure(request, error_builder.string_view());
  197. if (error_callback)
  198. error_callback(error_builder.to_string(), {});
  199. return;
  200. }
  201. log_success(request);
  202. success_callback(payload, response_headers, status_code);
  203. deferred_invoke([this, &protocol_request] {
  204. m_active_requests.remove(protocol_request);
  205. });
  206. };
  207. protocol_request->set_should_buffer_all_input(true);
  208. protocol_request->on_certificate_requested = []() -> Protocol::Request::CertificateAndKey {
  209. return {};
  210. };
  211. ++m_pending_loads;
  212. if (on_load_counter_change)
  213. on_load_counter_change();
  214. return;
  215. }
  216. auto not_implemented_error = String::formatted("Protocol not implemented: {}", url.protocol());
  217. log_failure(request, not_implemented_error);
  218. if (error_callback)
  219. error_callback(not_implemented_error, {});
  220. }
  221. void ResourceLoader::load(const AK::URL& url, Function<void(ReadonlyBytes, const HashMap<String, String, CaseInsensitiveStringTraits>& response_headers, Optional<u32> status_code)> success_callback, Function<void(const String&, Optional<u32> status_code)> error_callback)
  222. {
  223. LoadRequest request;
  224. request.set_url(url);
  225. load(request, move(success_callback), move(error_callback));
  226. }
  227. bool ResourceLoader::is_port_blocked(int port)
  228. {
  229. int ports[] { 1, 7, 9, 11, 13, 15, 17, 19, 20, 21, 22, 23, 25, 37, 42,
  230. 43, 53, 77, 79, 87, 95, 101, 102, 103, 104, 109, 110, 111, 113,
  231. 115, 117, 119, 123, 135, 139, 143, 179, 389, 465, 512, 513, 514,
  232. 515, 526, 530, 531, 532, 540, 556, 563, 587, 601, 636, 993, 995,
  233. 2049, 3659, 4045, 6000, 6379, 6665, 6666, 6667, 6668, 6669, 9000 };
  234. for (auto blocked_port : ports)
  235. if (port == blocked_port)
  236. return true;
  237. return false;
  238. }
  239. void ResourceLoader::clear_cache()
  240. {
  241. dbgln_if(CACHE_DEBUG, "Clearing {} items from ResourceLoader cache", s_resource_cache.size());
  242. s_resource_cache.clear();
  243. }
  244. }