ResourceLoader.cpp 9.8 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272
  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 ResourceLoader* s_the;
  22. if (!s_the)
  23. s_the = &ResourceLoader::construct().leak_ref();
  24. return *s_the;
  25. }
  26. ResourceLoader::ResourceLoader()
  27. : m_protocol_client(Protocol::RequestClient::construct())
  28. , m_user_agent(default_user_agent)
  29. {
  30. }
  31. 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)
  32. {
  33. Core::EventLoop loop;
  34. load(
  35. request,
  36. [&](auto data, auto& response_headers, auto status_code) {
  37. success_callback(data, response_headers, status_code);
  38. loop.quit(0);
  39. },
  40. [&](auto& string, auto status_code) {
  41. if (error_callback)
  42. error_callback(string, status_code);
  43. loop.quit(0);
  44. });
  45. loop.exec();
  46. }
  47. void ResourceLoader::prefetch_dns(AK::URL const& url)
  48. {
  49. m_protocol_client->ensure_connection(url, RequestServer::CacheLevel::ResolveOnly);
  50. }
  51. void ResourceLoader::preconnect(AK::URL const& url)
  52. {
  53. m_protocol_client->ensure_connection(url, RequestServer::CacheLevel::CreateConnection);
  54. }
  55. static HashMap<LoadRequest, NonnullRefPtr<Resource>> s_resource_cache;
  56. RefPtr<Resource> ResourceLoader::load_resource(Resource::Type type, LoadRequest& request)
  57. {
  58. if (!request.is_valid())
  59. return nullptr;
  60. bool use_cache = request.url().protocol() != "file";
  61. if (use_cache) {
  62. auto it = s_resource_cache.find(request);
  63. if (it != s_resource_cache.end()) {
  64. if (it->value->type() != type) {
  65. dbgln("FIXME: Not using cached resource for {} since there's a type mismatch.", request.url());
  66. } else {
  67. dbgln_if(CACHE_DEBUG, "Reusing cached resource for: {}", request.url());
  68. return it->value;
  69. }
  70. }
  71. }
  72. auto resource = Resource::create({}, type, request);
  73. if (use_cache)
  74. s_resource_cache.set(request, resource);
  75. load(
  76. request,
  77. [=](auto data, auto& headers, auto status_code) {
  78. const_cast<Resource&>(*resource).did_load({}, data, headers, status_code);
  79. },
  80. [=](auto& error, auto status_code) {
  81. const_cast<Resource&>(*resource).did_fail({}, error, status_code);
  82. });
  83. return resource;
  84. }
  85. static String sanitized_url_for_logging(AK::URL const& url)
  86. {
  87. if (url.protocol() == "data"sv)
  88. return String::formatted("[data URL, mime-type={}, size={}]", url.data_mime_type(), url.data_payload().length());
  89. return url.to_string();
  90. }
  91. 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)
  92. {
  93. auto& url = request.url();
  94. request.start_timer();
  95. dbgln("ResourceLoader: Starting load of: \"{}\"", sanitized_url_for_logging(url));
  96. const auto log_success = [](const auto& request) {
  97. auto& url = request.url();
  98. auto load_time_ms = request.load_time().to_milliseconds();
  99. dbgln("ResourceLoader: Finished load of: \"{}\", Duration: {}ms", sanitized_url_for_logging(url), load_time_ms);
  100. };
  101. const auto log_failure = [](const auto& request, const auto error_message) {
  102. auto& url = request.url();
  103. auto load_time_ms = request.load_time().to_milliseconds();
  104. dbgln("ResourceLoader: Failed load of: \"{}\", \033[32;1mError: {}\033[0m, Duration: {}ms", sanitized_url_for_logging(url), 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.has_value()) {
  133. auto error_message = "Base64 data contains an invalid character"sv;
  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. HashMap<String, String> headers;
  167. headers.set("User-Agent", m_user_agent);
  168. headers.set("Accept-Encoding", "gzip, deflate");
  169. for (auto& it : request.headers()) {
  170. headers.set(it.key, it.value);
  171. }
  172. auto protocol_request = protocol_client().start_request(request.method(), url, headers, request.body());
  173. if (!protocol_request) {
  174. auto start_request_failure_msg = "Failed to initiate load"sv;
  175. log_failure(request, start_request_failure_msg);
  176. if (error_callback)
  177. error_callback(start_request_failure_msg, {});
  178. return;
  179. }
  180. m_active_requests.set(*protocol_request);
  181. 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) {
  182. --m_pending_loads;
  183. if (on_load_counter_change)
  184. on_load_counter_change();
  185. if (!success) {
  186. auto http_load_failure_msg = "HTTP load failed"sv;
  187. log_failure(request, http_load_failure_msg);
  188. if (error_callback)
  189. error_callback(http_load_failure_msg, {});
  190. return;
  191. }
  192. log_success(request);
  193. success_callback(payload, response_headers, status_code);
  194. deferred_invoke([this, &protocol_request] {
  195. m_active_requests.remove(protocol_request);
  196. });
  197. };
  198. protocol_request->set_should_buffer_all_input(true);
  199. protocol_request->on_certificate_requested = []() -> Protocol::Request::CertificateAndKey {
  200. return {};
  201. };
  202. ++m_pending_loads;
  203. if (on_load_counter_change)
  204. on_load_counter_change();
  205. return;
  206. }
  207. auto not_implemented_error = String::formatted("Protocol not implemented: {}", url.protocol());
  208. log_failure(request, not_implemented_error);
  209. if (error_callback)
  210. error_callback(not_implemented_error, {});
  211. }
  212. 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)
  213. {
  214. LoadRequest request;
  215. request.set_url(url);
  216. load(request, move(success_callback), move(error_callback));
  217. }
  218. bool ResourceLoader::is_port_blocked(int port)
  219. {
  220. int ports[] { 1, 7, 9, 11, 13, 15, 17, 19, 20, 21, 22, 23, 25, 37, 42,
  221. 43, 53, 77, 79, 87, 95, 101, 102, 103, 104, 109, 110, 111, 113,
  222. 115, 117, 119, 123, 135, 139, 143, 179, 389, 465, 512, 513, 514,
  223. 515, 526, 530, 531, 532, 540, 556, 563, 587, 601, 636, 993, 995,
  224. 2049, 3659, 4045, 6000, 6379, 6665, 6666, 6667, 6668, 6669, 9000 };
  225. for (auto blocked_port : ports)
  226. if (port == blocked_port)
  227. return true;
  228. return false;
  229. }
  230. void ResourceLoader::clear_cache()
  231. {
  232. dbgln_if(CACHE_DEBUG, "Clearing {} items from ResourceLoader cache", s_resource_cache.size());
  233. s_resource_cache.clear();
  234. }
  235. }