ResourceLoader.cpp 9.3 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254
  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. static HashMap<LoadRequest, NonnullRefPtr<Resource>> s_resource_cache;
  48. RefPtr<Resource> ResourceLoader::load_resource(Resource::Type type, LoadRequest& request)
  49. {
  50. if (!request.is_valid())
  51. return nullptr;
  52. bool use_cache = request.url().protocol() != "file";
  53. if (use_cache) {
  54. auto it = s_resource_cache.find(request);
  55. if (it != s_resource_cache.end()) {
  56. if (it->value->type() != type) {
  57. dbgln("FIXME: Not using cached resource for {} since there's a type mismatch.", request.url());
  58. } else {
  59. dbgln_if(CACHE_DEBUG, "Reusing cached resource for: {}", request.url());
  60. return it->value;
  61. }
  62. }
  63. }
  64. auto resource = Resource::create({}, type, request);
  65. if (use_cache)
  66. s_resource_cache.set(request, resource);
  67. load(
  68. request,
  69. [=](auto data, auto& headers, auto status_code) {
  70. const_cast<Resource&>(*resource).did_load({}, data, headers, status_code);
  71. },
  72. [=](auto& error, auto status_code) {
  73. const_cast<Resource&>(*resource).did_fail({}, error, status_code);
  74. });
  75. return resource;
  76. }
  77. static String sanitized_url_for_logging(AK::URL const& url)
  78. {
  79. if (url.protocol() == "data"sv)
  80. return String::formatted("[data URL, mime-type={}, size={}]", url.data_mime_type(), url.data_payload().length());
  81. return url.to_string();
  82. }
  83. 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)
  84. {
  85. auto& url = request.url();
  86. request.start_timer();
  87. dbgln("ResourceLoader: Starting load of: \"{}\"", sanitized_url_for_logging(url));
  88. const auto log_success = [](const auto& request) {
  89. auto& url = request.url();
  90. auto load_time_ms = request.load_time().to_milliseconds();
  91. dbgln("ResourceLoader: Finished load of: \"{}\", Duration: {}ms", sanitized_url_for_logging(url), load_time_ms);
  92. };
  93. const auto log_failure = [](const auto& request, const auto error_message) {
  94. auto& url = request.url();
  95. auto load_time_ms = request.load_time().to_milliseconds();
  96. dbgln("ResourceLoader: Failed load of: \"{}\", \033[32;1mError: {}\033[0m, Duration: {}ms", sanitized_url_for_logging(url), error_message, load_time_ms);
  97. };
  98. if (is_port_blocked(url.port_or_default())) {
  99. log_failure(request, String::formatted("The port #{} is blocked", url.port_or_default()));
  100. return;
  101. }
  102. if (ContentFilter::the().is_filtered(url)) {
  103. auto filter_message = "URL was filtered"sv;
  104. log_failure(request, filter_message);
  105. error_callback(filter_message, {});
  106. return;
  107. }
  108. if (url.protocol() == "about") {
  109. dbgln_if(SPAM_DEBUG, "Loading about: URL {}", url);
  110. deferred_invoke([success_callback = move(success_callback)] {
  111. success_callback(String::empty().to_byte_buffer(), {}, {});
  112. });
  113. log_success(request);
  114. return;
  115. }
  116. if (url.protocol() == "data") {
  117. dbgln_if(SPAM_DEBUG, "ResourceLoader loading a data URL with mime-type: '{}', base64={}, payload='{}'",
  118. url.data_mime_type(),
  119. url.data_payload_is_base64(),
  120. url.data_payload());
  121. ByteBuffer data;
  122. if (url.data_payload_is_base64())
  123. data = decode_base64(url.data_payload());
  124. else
  125. data = url.data_payload().to_byte_buffer();
  126. deferred_invoke([data = move(data), success_callback = move(success_callback)] {
  127. success_callback(data, {}, {});
  128. });
  129. log_success(request);
  130. return;
  131. }
  132. if (url.protocol() == "file") {
  133. auto file_result = Core::File::open(url.path(), Core::OpenMode::ReadOnly);
  134. if (file_result.is_error()) {
  135. auto& error = file_result.error();
  136. log_failure(request, error.string());
  137. if (error_callback)
  138. error_callback(error.string(), error.error());
  139. return;
  140. }
  141. auto file = file_result.release_value();
  142. auto data = file->read_all();
  143. deferred_invoke([data = move(data), success_callback = move(success_callback)] {
  144. success_callback(data, {}, {});
  145. });
  146. log_success(request);
  147. return;
  148. }
  149. if (url.protocol() == "http" || url.protocol() == "https" || url.protocol() == "gemini") {
  150. HashMap<String, String> headers;
  151. headers.set("User-Agent", m_user_agent);
  152. headers.set("Accept-Encoding", "gzip, deflate");
  153. for (auto& it : request.headers()) {
  154. headers.set(it.key, it.value);
  155. }
  156. auto protocol_request = protocol_client().start_request(request.method(), url, headers, request.body());
  157. if (!protocol_request) {
  158. auto start_request_failure_msg = "Failed to initiate load"sv;
  159. log_failure(request, start_request_failure_msg);
  160. if (error_callback)
  161. error_callback(start_request_failure_msg, {});
  162. return;
  163. }
  164. protocol_request->on_buffered_request_finish = [this, success_callback = move(success_callback), error_callback = move(error_callback), log_success, log_failure, request, protocol_request](bool success, auto, auto& response_headers, auto status_code, ReadonlyBytes payload) {
  165. --m_pending_loads;
  166. if (on_load_counter_change)
  167. on_load_counter_change();
  168. if (!success) {
  169. auto http_load_failure_msg = "HTTP load failed"sv;
  170. log_failure(request, http_load_failure_msg);
  171. if (error_callback)
  172. error_callback(http_load_failure_msg, {});
  173. return;
  174. }
  175. deferred_invoke([protocol_request] {
  176. // Clear circular reference of `protocol_request` captured by copy
  177. const_cast<Protocol::Request&>(*protocol_request).on_buffered_request_finish = nullptr;
  178. });
  179. success_callback(payload, response_headers, status_code);
  180. log_success(request);
  181. };
  182. protocol_request->set_should_buffer_all_input(true);
  183. protocol_request->on_certificate_requested = []() -> Protocol::Request::CertificateAndKey {
  184. return {};
  185. };
  186. ++m_pending_loads;
  187. if (on_load_counter_change)
  188. on_load_counter_change();
  189. return;
  190. }
  191. auto not_implemented_error = String::formatted("Protocol not implemented: {}", url.protocol());
  192. log_failure(request, not_implemented_error);
  193. if (error_callback)
  194. error_callback(not_implemented_error, {});
  195. }
  196. 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)
  197. {
  198. LoadRequest request;
  199. request.set_url(url);
  200. load(request, move(success_callback), move(error_callback));
  201. }
  202. bool ResourceLoader::is_port_blocked(int port)
  203. {
  204. int ports[] { 1, 7, 9, 11, 13, 15, 17, 19, 20, 21, 22, 23, 25, 37, 42,
  205. 43, 53, 77, 79, 87, 95, 101, 102, 103, 104, 109, 110, 111, 113,
  206. 115, 117, 119, 123, 135, 139, 143, 179, 389, 465, 512, 513, 514,
  207. 515, 526, 530, 531, 532, 540, 556, 563, 587, 601, 636, 993, 995,
  208. 2049, 3659, 4045, 6000, 6379, 6665, 6666, 6667, 6668, 6669, 9000 };
  209. for (auto blocked_port : ports)
  210. if (port == blocked_port)
  211. return true;
  212. return false;
  213. }
  214. void ResourceLoader::clear_cache()
  215. {
  216. dbgln_if(CACHE_DEBUG, "Clearing {} items from ResourceLoader cache", s_resource_cache.size());
  217. s_resource_cache.clear();
  218. }
  219. }