worker.js 17 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551
  1. /**
  2. * @licstart The following is the entire license notice for the
  3. * Javascript code in this page
  4. *
  5. * Copyright 2020 Mozilla Foundation
  6. *
  7. * Licensed under the Apache License, Version 2.0 (the "License");
  8. * you may not use this file except in compliance with the License.
  9. * You may obtain a copy of the License at
  10. *
  11. * http://www.apache.org/licenses/LICENSE-2.0
  12. *
  13. * Unless required by applicable law or agreed to in writing, software
  14. * distributed under the License is distributed on an "AS IS" BASIS,
  15. * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  16. * See the License for the specific language governing permissions and
  17. * limitations under the License.
  18. *
  19. * @licend The above is the entire license notice for the
  20. * Javascript code in this page
  21. */
  22. "use strict";
  23. Object.defineProperty(exports, "__esModule", {
  24. value: true
  25. });
  26. exports.WorkerMessageHandler = exports.WorkerTask = void 0;
  27. var _util = require("../shared/util.js");
  28. var _primitives = require("./primitives.js");
  29. var _pdf_manager = require("./pdf_manager.js");
  30. var _is_node = require("../shared/is_node.js");
  31. var _message_handler = require("../shared/message_handler.js");
  32. var _worker_stream = require("./worker_stream.js");
  33. var _core_utils = require("./core_utils.js");
  34. var WorkerTask = function WorkerTaskClosure() {
  35. function WorkerTask(name) {
  36. this.name = name;
  37. this.terminated = false;
  38. this._capability = (0, _util.createPromiseCapability)();
  39. }
  40. WorkerTask.prototype = {
  41. get finished() {
  42. return this._capability.promise;
  43. },
  44. finish() {
  45. this._capability.resolve();
  46. },
  47. terminate() {
  48. this.terminated = true;
  49. },
  50. ensureNotTerminated() {
  51. if (this.terminated) {
  52. throw new Error("Worker task was terminated");
  53. }
  54. }
  55. };
  56. return WorkerTask;
  57. }();
  58. exports.WorkerTask = WorkerTask;
  59. var WorkerMessageHandler = {
  60. setup(handler, port) {
  61. var testMessageProcessed = false;
  62. handler.on("test", function wphSetupTest(data) {
  63. if (testMessageProcessed) {
  64. return;
  65. }
  66. testMessageProcessed = true;
  67. if (!(data instanceof Uint8Array)) {
  68. handler.send("test", null);
  69. return;
  70. }
  71. const supportTransfers = data[0] === 255;
  72. handler.postMessageTransfers = supportTransfers;
  73. handler.send("test", {
  74. supportTransfers
  75. });
  76. });
  77. handler.on("configure", function wphConfigure(data) {
  78. (0, _util.setVerbosityLevel)(data.verbosity);
  79. });
  80. handler.on("GetDocRequest", function wphSetupDoc(data) {
  81. return WorkerMessageHandler.createDocumentHandler(data, port);
  82. });
  83. },
  84. createDocumentHandler(docParams, port) {
  85. var pdfManager;
  86. var terminated = false;
  87. var cancelXHRs = null;
  88. var WorkerTasks = [];
  89. const verbosity = (0, _util.getVerbosityLevel)();
  90. const apiVersion = docParams.apiVersion;
  91. const workerVersion = '2.4.456';
  92. if (apiVersion !== workerVersion) {
  93. throw new Error(`The API version "${apiVersion}" does not match ` + `the Worker version "${workerVersion}".`);
  94. }
  95. const enumerableProperties = [];
  96. for (const property in []) {
  97. enumerableProperties.push(property);
  98. }
  99. if (enumerableProperties.length) {
  100. throw new Error("The `Array.prototype` contains unexpected enumerable properties: " + enumerableProperties.join(", ") + "; thus breaking e.g. `for...in` iteration of `Array`s.");
  101. }
  102. var docId = docParams.docId;
  103. var docBaseUrl = docParams.docBaseUrl;
  104. var workerHandlerName = docParams.docId + "_worker";
  105. var handler = new _message_handler.MessageHandler(workerHandlerName, docId, port);
  106. handler.postMessageTransfers = docParams.postMessageTransfers;
  107. function ensureNotTerminated() {
  108. if (terminated) {
  109. throw new Error("Worker was terminated");
  110. }
  111. }
  112. function startWorkerTask(task) {
  113. WorkerTasks.push(task);
  114. }
  115. function finishWorkerTask(task) {
  116. task.finish();
  117. var i = WorkerTasks.indexOf(task);
  118. WorkerTasks.splice(i, 1);
  119. }
  120. async function loadDocument(recoveryMode) {
  121. await pdfManager.ensureDoc("checkHeader");
  122. await pdfManager.ensureDoc("parseStartXRef");
  123. await pdfManager.ensureDoc("parse", [recoveryMode]);
  124. if (!recoveryMode) {
  125. await pdfManager.ensureDoc("checkFirstPage");
  126. }
  127. const [numPages, fingerprint] = await Promise.all([pdfManager.ensureDoc("numPages"), pdfManager.ensureDoc("fingerprint")]);
  128. return {
  129. numPages,
  130. fingerprint
  131. };
  132. }
  133. function getPdfManager(data, evaluatorOptions) {
  134. var pdfManagerCapability = (0, _util.createPromiseCapability)();
  135. var pdfManager;
  136. var source = data.source;
  137. if (source.data) {
  138. try {
  139. pdfManager = new _pdf_manager.LocalPdfManager(docId, source.data, source.password, evaluatorOptions, docBaseUrl);
  140. pdfManagerCapability.resolve(pdfManager);
  141. } catch (ex) {
  142. pdfManagerCapability.reject(ex);
  143. }
  144. return pdfManagerCapability.promise;
  145. }
  146. var pdfStream,
  147. cachedChunks = [];
  148. try {
  149. pdfStream = new _worker_stream.PDFWorkerStream(handler);
  150. } catch (ex) {
  151. pdfManagerCapability.reject(ex);
  152. return pdfManagerCapability.promise;
  153. }
  154. var fullRequest = pdfStream.getFullReader();
  155. fullRequest.headersReady.then(function () {
  156. if (!fullRequest.isRangeSupported) {
  157. return;
  158. }
  159. var disableAutoFetch = source.disableAutoFetch || fullRequest.isStreamingSupported;
  160. pdfManager = new _pdf_manager.NetworkPdfManager(docId, pdfStream, {
  161. msgHandler: handler,
  162. password: source.password,
  163. length: fullRequest.contentLength,
  164. disableAutoFetch,
  165. rangeChunkSize: source.rangeChunkSize
  166. }, evaluatorOptions, docBaseUrl);
  167. for (let i = 0; i < cachedChunks.length; i++) {
  168. pdfManager.sendProgressiveData(cachedChunks[i]);
  169. }
  170. cachedChunks = [];
  171. pdfManagerCapability.resolve(pdfManager);
  172. cancelXHRs = null;
  173. }).catch(function (reason) {
  174. pdfManagerCapability.reject(reason);
  175. cancelXHRs = null;
  176. });
  177. var loaded = 0;
  178. var flushChunks = function () {
  179. var pdfFile = (0, _util.arraysToBytes)(cachedChunks);
  180. if (source.length && pdfFile.length !== source.length) {
  181. (0, _util.warn)("reported HTTP length is different from actual");
  182. }
  183. try {
  184. pdfManager = new _pdf_manager.LocalPdfManager(docId, pdfFile, source.password, evaluatorOptions, docBaseUrl);
  185. pdfManagerCapability.resolve(pdfManager);
  186. } catch (ex) {
  187. pdfManagerCapability.reject(ex);
  188. }
  189. cachedChunks = [];
  190. };
  191. var readPromise = new Promise(function (resolve, reject) {
  192. var readChunk = function (chunk) {
  193. try {
  194. ensureNotTerminated();
  195. if (chunk.done) {
  196. if (!pdfManager) {
  197. flushChunks();
  198. }
  199. cancelXHRs = null;
  200. return;
  201. }
  202. var data = chunk.value;
  203. loaded += (0, _util.arrayByteLength)(data);
  204. if (!fullRequest.isStreamingSupported) {
  205. handler.send("DocProgress", {
  206. loaded,
  207. total: Math.max(loaded, fullRequest.contentLength || 0)
  208. });
  209. }
  210. if (pdfManager) {
  211. pdfManager.sendProgressiveData(data);
  212. } else {
  213. cachedChunks.push(data);
  214. }
  215. fullRequest.read().then(readChunk, reject);
  216. } catch (e) {
  217. reject(e);
  218. }
  219. };
  220. fullRequest.read().then(readChunk, reject);
  221. });
  222. readPromise.catch(function (e) {
  223. pdfManagerCapability.reject(e);
  224. cancelXHRs = null;
  225. });
  226. cancelXHRs = function (reason) {
  227. pdfStream.cancelAllRequests(reason);
  228. };
  229. return pdfManagerCapability.promise;
  230. }
  231. function setupDoc(data) {
  232. function onSuccess(doc) {
  233. ensureNotTerminated();
  234. handler.send("GetDoc", {
  235. pdfInfo: doc
  236. });
  237. }
  238. function onFailure(ex) {
  239. ensureNotTerminated();
  240. if (ex instanceof _util.PasswordException) {
  241. var task = new WorkerTask(`PasswordException: response ${ex.code}`);
  242. startWorkerTask(task);
  243. handler.sendWithPromise("PasswordRequest", ex).then(function (data) {
  244. finishWorkerTask(task);
  245. pdfManager.updatePassword(data.password);
  246. pdfManagerReady();
  247. }).catch(function () {
  248. finishWorkerTask(task);
  249. handler.send("DocException", ex);
  250. });
  251. } else if (ex instanceof _util.InvalidPDFException || ex instanceof _util.MissingPDFException || ex instanceof _util.UnexpectedResponseException || ex instanceof _util.UnknownErrorException) {
  252. handler.send("DocException", ex);
  253. } else {
  254. handler.send("DocException", new _util.UnknownErrorException(ex.message, ex.toString()));
  255. }
  256. }
  257. function pdfManagerReady() {
  258. ensureNotTerminated();
  259. loadDocument(false).then(onSuccess, function loadFailure(ex) {
  260. ensureNotTerminated();
  261. if (!(ex instanceof _core_utils.XRefParseException)) {
  262. onFailure(ex);
  263. return;
  264. }
  265. pdfManager.requestLoadedStream();
  266. pdfManager.onLoadedStream().then(function () {
  267. ensureNotTerminated();
  268. loadDocument(true).then(onSuccess, onFailure);
  269. });
  270. }, onFailure);
  271. }
  272. ensureNotTerminated();
  273. var evaluatorOptions = {
  274. forceDataSchema: data.disableCreateObjectURL,
  275. maxImageSize: data.maxImageSize,
  276. disableFontFace: data.disableFontFace,
  277. nativeImageDecoderSupport: data.nativeImageDecoderSupport,
  278. ignoreErrors: data.ignoreErrors,
  279. isEvalSupported: data.isEvalSupported
  280. };
  281. getPdfManager(data, evaluatorOptions).then(function (newPdfManager) {
  282. if (terminated) {
  283. newPdfManager.terminate(new _util.AbortException("Worker was terminated."));
  284. throw new Error("Worker was terminated");
  285. }
  286. pdfManager = newPdfManager;
  287. pdfManager.onLoadedStream().then(function (stream) {
  288. handler.send("DataLoaded", {
  289. length: stream.bytes.byteLength
  290. });
  291. });
  292. }).then(pdfManagerReady, onFailure);
  293. }
  294. handler.on("GetPage", function wphSetupGetPage(data) {
  295. return pdfManager.getPage(data.pageIndex).then(function (page) {
  296. return Promise.all([pdfManager.ensure(page, "rotate"), pdfManager.ensure(page, "ref"), pdfManager.ensure(page, "userUnit"), pdfManager.ensure(page, "view")]).then(function ([rotate, ref, userUnit, view]) {
  297. return {
  298. rotate,
  299. ref,
  300. userUnit,
  301. view
  302. };
  303. });
  304. });
  305. });
  306. handler.on("GetPageIndex", function wphSetupGetPageIndex(data) {
  307. var ref = _primitives.Ref.get(data.ref.num, data.ref.gen);
  308. var catalog = pdfManager.pdfDocument.catalog;
  309. return catalog.getPageIndex(ref);
  310. });
  311. handler.on("GetDestinations", function wphSetupGetDestinations(data) {
  312. return pdfManager.ensureCatalog("destinations");
  313. });
  314. handler.on("GetDestination", function wphSetupGetDestination(data) {
  315. return pdfManager.ensureCatalog("getDestination", [data.id]);
  316. });
  317. handler.on("GetPageLabels", function wphSetupGetPageLabels(data) {
  318. return pdfManager.ensureCatalog("pageLabels");
  319. });
  320. handler.on("GetPageLayout", function wphSetupGetPageLayout(data) {
  321. return pdfManager.ensureCatalog("pageLayout");
  322. });
  323. handler.on("GetPageMode", function wphSetupGetPageMode(data) {
  324. return pdfManager.ensureCatalog("pageMode");
  325. });
  326. handler.on("GetViewerPreferences", function (data) {
  327. return pdfManager.ensureCatalog("viewerPreferences");
  328. });
  329. handler.on("GetOpenAction", function (data) {
  330. return pdfManager.ensureCatalog("openAction");
  331. });
  332. handler.on("GetAttachments", function wphSetupGetAttachments(data) {
  333. return pdfManager.ensureCatalog("attachments");
  334. });
  335. handler.on("GetJavaScript", function wphSetupGetJavaScript(data) {
  336. return pdfManager.ensureCatalog("javaScript");
  337. });
  338. handler.on("GetOutline", function wphSetupGetOutline(data) {
  339. return pdfManager.ensureCatalog("documentOutline");
  340. });
  341. handler.on("GetPermissions", function (data) {
  342. return pdfManager.ensureCatalog("permissions");
  343. });
  344. handler.on("GetMetadata", function wphSetupGetMetadata(data) {
  345. return Promise.all([pdfManager.ensureDoc("documentInfo"), pdfManager.ensureCatalog("metadata")]);
  346. });
  347. handler.on("GetData", function wphSetupGetData(data) {
  348. pdfManager.requestLoadedStream();
  349. return pdfManager.onLoadedStream().then(function (stream) {
  350. return stream.bytes;
  351. });
  352. });
  353. handler.on("GetStats", function wphSetupGetStats(data) {
  354. return pdfManager.pdfDocument.xref.stats;
  355. });
  356. handler.on("GetAnnotations", function ({
  357. pageIndex,
  358. intent
  359. }) {
  360. return pdfManager.getPage(pageIndex).then(function (page) {
  361. return page.getAnnotationsData(intent);
  362. });
  363. });
  364. handler.on("GetOperatorList", function wphSetupRenderPage(data, sink) {
  365. var pageIndex = data.pageIndex;
  366. pdfManager.getPage(pageIndex).then(function (page) {
  367. var task = new WorkerTask(`GetOperatorList: page ${pageIndex}`);
  368. startWorkerTask(task);
  369. const start = verbosity >= _util.VerbosityLevel.INFOS ? Date.now() : 0;
  370. page.getOperatorList({
  371. handler,
  372. sink,
  373. task,
  374. intent: data.intent,
  375. renderInteractiveForms: data.renderInteractiveForms
  376. }).then(function (operatorListInfo) {
  377. finishWorkerTask(task);
  378. if (start) {
  379. (0, _util.info)(`page=${pageIndex + 1} - getOperatorList: time=` + `${Date.now() - start}ms, len=${operatorListInfo.length}`);
  380. }
  381. sink.close();
  382. }, function (reason) {
  383. finishWorkerTask(task);
  384. if (task.terminated) {
  385. return;
  386. }
  387. handler.send("UnsupportedFeature", {
  388. featureId: _util.UNSUPPORTED_FEATURES.unknown
  389. });
  390. sink.error(reason);
  391. });
  392. });
  393. }, this);
  394. handler.on("GetTextContent", function wphExtractText(data, sink) {
  395. var pageIndex = data.pageIndex;
  396. sink.onPull = function (desiredSize) {};
  397. sink.onCancel = function (reason) {};
  398. pdfManager.getPage(pageIndex).then(function (page) {
  399. var task = new WorkerTask("GetTextContent: page " + pageIndex);
  400. startWorkerTask(task);
  401. const start = verbosity >= _util.VerbosityLevel.INFOS ? Date.now() : 0;
  402. page.extractTextContent({
  403. handler,
  404. task,
  405. sink,
  406. normalizeWhitespace: data.normalizeWhitespace,
  407. combineTextItems: data.combineTextItems
  408. }).then(function () {
  409. finishWorkerTask(task);
  410. if (start) {
  411. (0, _util.info)(`page=${pageIndex + 1} - getTextContent: time=` + `${Date.now() - start}ms`);
  412. }
  413. sink.close();
  414. }, function (reason) {
  415. finishWorkerTask(task);
  416. if (task.terminated) {
  417. return;
  418. }
  419. sink.error(reason);
  420. });
  421. });
  422. });
  423. handler.on("FontFallback", function (data) {
  424. return pdfManager.fontFallback(data.id, handler);
  425. });
  426. handler.on("Cleanup", function wphCleanup(data) {
  427. return pdfManager.cleanup();
  428. });
  429. handler.on("Terminate", function wphTerminate(data) {
  430. terminated = true;
  431. const waitOn = [];
  432. if (pdfManager) {
  433. pdfManager.terminate(new _util.AbortException("Worker was terminated."));
  434. const cleanupPromise = pdfManager.cleanup();
  435. waitOn.push(cleanupPromise);
  436. pdfManager = null;
  437. } else {
  438. (0, _primitives.clearPrimitiveCaches)();
  439. }
  440. if (cancelXHRs) {
  441. cancelXHRs(new _util.AbortException("Worker was terminated."));
  442. }
  443. WorkerTasks.forEach(function (task) {
  444. waitOn.push(task.finished);
  445. task.terminate();
  446. });
  447. return Promise.all(waitOn).then(function () {
  448. handler.destroy();
  449. handler = null;
  450. });
  451. });
  452. handler.on("Ready", function wphReady(data) {
  453. setupDoc(docParams);
  454. docParams = null;
  455. });
  456. return workerHandlerName;
  457. },
  458. initializeFromPort(port) {
  459. var handler = new _message_handler.MessageHandler("worker", "main", port);
  460. WorkerMessageHandler.setup(handler, port);
  461. handler.send("ready", null);
  462. }
  463. };
  464. exports.WorkerMessageHandler = WorkerMessageHandler;
  465. function isMessagePort(maybePort) {
  466. return typeof maybePort.postMessage === "function" && "onmessage" in maybePort;
  467. }
  468. if (typeof window === "undefined" && !_is_node.isNodeJS && typeof self !== "undefined" && isMessagePort(self)) {
  469. WorkerMessageHandler.initializeFromPort(self);
  470. }