worker.js 19 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623
  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 _writer = require("./writer.js");
  31. var _is_node = require("../shared/is_node.js");
  32. var _message_handler = require("../shared/message_handler.js");
  33. var _worker_stream = require("./worker_stream.js");
  34. var _core_utils = require("./core_utils.js");
  35. class WorkerTask {
  36. constructor(name) {
  37. this.name = name;
  38. this.terminated = false;
  39. this._capability = (0, _util.createPromiseCapability)();
  40. }
  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. exports.WorkerTask = WorkerTask;
  57. class WorkerMessageHandler {
  58. static setup(handler, port) {
  59. var testMessageProcessed = false;
  60. handler.on("test", function wphSetupTest(data) {
  61. if (testMessageProcessed) {
  62. return;
  63. }
  64. testMessageProcessed = true;
  65. if (!(data instanceof Uint8Array)) {
  66. handler.send("test", null);
  67. return;
  68. }
  69. const supportTransfers = data[0] === 255;
  70. handler.postMessageTransfers = supportTransfers;
  71. handler.send("test", {
  72. supportTransfers
  73. });
  74. });
  75. handler.on("configure", function wphConfigure(data) {
  76. (0, _util.setVerbosityLevel)(data.verbosity);
  77. });
  78. handler.on("GetDocRequest", function wphSetupDoc(data) {
  79. return WorkerMessageHandler.createDocumentHandler(data, port);
  80. });
  81. }
  82. static createDocumentHandler(docParams, port) {
  83. var pdfManager;
  84. var terminated = false;
  85. var cancelXHRs = null;
  86. var WorkerTasks = [];
  87. const verbosity = (0, _util.getVerbosityLevel)();
  88. const apiVersion = docParams.apiVersion;
  89. const workerVersion = '2.6.347';
  90. if (apiVersion !== workerVersion) {
  91. throw new Error(`The API version "${apiVersion}" does not match ` + `the Worker version "${workerVersion}".`);
  92. }
  93. const enumerableProperties = [];
  94. for (const property in []) {
  95. enumerableProperties.push(property);
  96. }
  97. if (enumerableProperties.length) {
  98. throw new Error("The `Array.prototype` contains unexpected enumerable properties: " + enumerableProperties.join(", ") + "; thus breaking e.g. `for...in` iteration of `Array`s.");
  99. }
  100. if (typeof ReadableStream === "undefined" || typeof Promise.allSettled === "undefined") {
  101. throw new Error("The browser/environment lacks native support for critical " + "functionality used by the PDF.js library (e.g. " + "`ReadableStream` and/or `Promise.allSettled`); " + "please use an ES5-compatible build instead.");
  102. }
  103. var docId = docParams.docId;
  104. var docBaseUrl = docParams.docBaseUrl;
  105. var workerHandlerName = docParams.docId + "_worker";
  106. var handler = new _message_handler.MessageHandler(workerHandlerName, docId, port);
  107. handler.postMessageTransfers = docParams.postMessageTransfers;
  108. function ensureNotTerminated() {
  109. if (terminated) {
  110. throw new Error("Worker was terminated");
  111. }
  112. }
  113. function startWorkerTask(task) {
  114. WorkerTasks.push(task);
  115. }
  116. function finishWorkerTask(task) {
  117. task.finish();
  118. var i = WorkerTasks.indexOf(task);
  119. WorkerTasks.splice(i, 1);
  120. }
  121. async function loadDocument(recoveryMode) {
  122. await pdfManager.ensureDoc("checkHeader");
  123. await pdfManager.ensureDoc("parseStartXRef");
  124. await pdfManager.ensureDoc("parse", [recoveryMode]);
  125. if (!recoveryMode) {
  126. await pdfManager.ensureDoc("checkFirstPage");
  127. }
  128. const [numPages, fingerprint] = await Promise.all([pdfManager.ensureDoc("numPages"), pdfManager.ensureDoc("fingerprint")]);
  129. return {
  130. numPages,
  131. fingerprint
  132. };
  133. }
  134. function getPdfManager(data, evaluatorOptions) {
  135. var pdfManagerCapability = (0, _util.createPromiseCapability)();
  136. let newPdfManager;
  137. var source = data.source;
  138. if (source.data) {
  139. try {
  140. newPdfManager = new _pdf_manager.LocalPdfManager(docId, source.data, source.password, evaluatorOptions, docBaseUrl);
  141. pdfManagerCapability.resolve(newPdfManager);
  142. } catch (ex) {
  143. pdfManagerCapability.reject(ex);
  144. }
  145. return pdfManagerCapability.promise;
  146. }
  147. var pdfStream,
  148. cachedChunks = [];
  149. try {
  150. pdfStream = new _worker_stream.PDFWorkerStream(handler);
  151. } catch (ex) {
  152. pdfManagerCapability.reject(ex);
  153. return pdfManagerCapability.promise;
  154. }
  155. var fullRequest = pdfStream.getFullReader();
  156. fullRequest.headersReady.then(function () {
  157. if (!fullRequest.isRangeSupported) {
  158. return;
  159. }
  160. var disableAutoFetch = source.disableAutoFetch || fullRequest.isStreamingSupported;
  161. newPdfManager = new _pdf_manager.NetworkPdfManager(docId, pdfStream, {
  162. msgHandler: handler,
  163. password: source.password,
  164. length: fullRequest.contentLength,
  165. disableAutoFetch,
  166. rangeChunkSize: source.rangeChunkSize
  167. }, evaluatorOptions, docBaseUrl);
  168. for (let i = 0; i < cachedChunks.length; i++) {
  169. newPdfManager.sendProgressiveData(cachedChunks[i]);
  170. }
  171. cachedChunks = [];
  172. pdfManagerCapability.resolve(newPdfManager);
  173. cancelXHRs = null;
  174. }).catch(function (reason) {
  175. pdfManagerCapability.reject(reason);
  176. cancelXHRs = null;
  177. });
  178. var loaded = 0;
  179. var flushChunks = function () {
  180. var pdfFile = (0, _util.arraysToBytes)(cachedChunks);
  181. if (source.length && pdfFile.length !== source.length) {
  182. (0, _util.warn)("reported HTTP length is different from actual");
  183. }
  184. try {
  185. newPdfManager = new _pdf_manager.LocalPdfManager(docId, pdfFile, source.password, evaluatorOptions, docBaseUrl);
  186. pdfManagerCapability.resolve(newPdfManager);
  187. } catch (ex) {
  188. pdfManagerCapability.reject(ex);
  189. }
  190. cachedChunks = [];
  191. };
  192. var readPromise = new Promise(function (resolve, reject) {
  193. var readChunk = function ({
  194. value,
  195. done
  196. }) {
  197. try {
  198. ensureNotTerminated();
  199. if (done) {
  200. if (!newPdfManager) {
  201. flushChunks();
  202. }
  203. cancelXHRs = null;
  204. return;
  205. }
  206. loaded += (0, _util.arrayByteLength)(value);
  207. if (!fullRequest.isStreamingSupported) {
  208. handler.send("DocProgress", {
  209. loaded,
  210. total: Math.max(loaded, fullRequest.contentLength || 0)
  211. });
  212. }
  213. if (newPdfManager) {
  214. newPdfManager.sendProgressiveData(value);
  215. } else {
  216. cachedChunks.push(value);
  217. }
  218. fullRequest.read().then(readChunk, reject);
  219. } catch (e) {
  220. reject(e);
  221. }
  222. };
  223. fullRequest.read().then(readChunk, reject);
  224. });
  225. readPromise.catch(function (e) {
  226. pdfManagerCapability.reject(e);
  227. cancelXHRs = null;
  228. });
  229. cancelXHRs = function (reason) {
  230. pdfStream.cancelAllRequests(reason);
  231. };
  232. return pdfManagerCapability.promise;
  233. }
  234. function setupDoc(data) {
  235. function onSuccess(doc) {
  236. ensureNotTerminated();
  237. handler.send("GetDoc", {
  238. pdfInfo: doc
  239. });
  240. }
  241. function onFailure(ex) {
  242. ensureNotTerminated();
  243. if (ex instanceof _util.PasswordException) {
  244. var task = new WorkerTask(`PasswordException: response ${ex.code}`);
  245. startWorkerTask(task);
  246. handler.sendWithPromise("PasswordRequest", ex).then(function ({
  247. password
  248. }) {
  249. finishWorkerTask(task);
  250. pdfManager.updatePassword(password);
  251. pdfManagerReady();
  252. }).catch(function () {
  253. finishWorkerTask(task);
  254. handler.send("DocException", ex);
  255. });
  256. } else if (ex instanceof _util.InvalidPDFException || ex instanceof _util.MissingPDFException || ex instanceof _util.UnexpectedResponseException || ex instanceof _util.UnknownErrorException) {
  257. handler.send("DocException", ex);
  258. } else {
  259. handler.send("DocException", new _util.UnknownErrorException(ex.message, ex.toString()));
  260. }
  261. }
  262. function pdfManagerReady() {
  263. ensureNotTerminated();
  264. loadDocument(false).then(onSuccess, function (reason) {
  265. ensureNotTerminated();
  266. if (!(reason instanceof _core_utils.XRefParseException)) {
  267. onFailure(reason);
  268. return;
  269. }
  270. pdfManager.requestLoadedStream();
  271. pdfManager.onLoadedStream().then(function () {
  272. ensureNotTerminated();
  273. loadDocument(true).then(onSuccess, onFailure);
  274. });
  275. });
  276. }
  277. ensureNotTerminated();
  278. var evaluatorOptions = {
  279. maxImageSize: data.maxImageSize,
  280. disableFontFace: data.disableFontFace,
  281. ignoreErrors: data.ignoreErrors,
  282. isEvalSupported: data.isEvalSupported,
  283. fontExtraProperties: data.fontExtraProperties
  284. };
  285. getPdfManager(data, evaluatorOptions).then(function (newPdfManager) {
  286. if (terminated) {
  287. newPdfManager.terminate(new _util.AbortException("Worker was terminated."));
  288. throw new Error("Worker was terminated");
  289. }
  290. pdfManager = newPdfManager;
  291. pdfManager.onLoadedStream().then(function (stream) {
  292. handler.send("DataLoaded", {
  293. length: stream.bytes.byteLength
  294. });
  295. });
  296. }).then(pdfManagerReady, onFailure);
  297. }
  298. handler.on("GetPage", function wphSetupGetPage(data) {
  299. return pdfManager.getPage(data.pageIndex).then(function (page) {
  300. 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]) {
  301. return {
  302. rotate,
  303. ref,
  304. userUnit,
  305. view
  306. };
  307. });
  308. });
  309. });
  310. handler.on("GetPageIndex", function wphSetupGetPageIndex({
  311. ref
  312. }) {
  313. const pageRef = _primitives.Ref.get(ref.num, ref.gen);
  314. return pdfManager.ensureCatalog("getPageIndex", [pageRef]);
  315. });
  316. handler.on("GetDestinations", function wphSetupGetDestinations(data) {
  317. return pdfManager.ensureCatalog("destinations");
  318. });
  319. handler.on("GetDestination", function wphSetupGetDestination(data) {
  320. return pdfManager.ensureCatalog("getDestination", [data.id]);
  321. });
  322. handler.on("GetPageLabels", function wphSetupGetPageLabels(data) {
  323. return pdfManager.ensureCatalog("pageLabels");
  324. });
  325. handler.on("GetPageLayout", function wphSetupGetPageLayout(data) {
  326. return pdfManager.ensureCatalog("pageLayout");
  327. });
  328. handler.on("GetPageMode", function wphSetupGetPageMode(data) {
  329. return pdfManager.ensureCatalog("pageMode");
  330. });
  331. handler.on("GetViewerPreferences", function (data) {
  332. return pdfManager.ensureCatalog("viewerPreferences");
  333. });
  334. handler.on("GetOpenAction", function (data) {
  335. return pdfManager.ensureCatalog("openAction");
  336. });
  337. handler.on("GetAttachments", function wphSetupGetAttachments(data) {
  338. return pdfManager.ensureCatalog("attachments");
  339. });
  340. handler.on("GetJavaScript", function wphSetupGetJavaScript(data) {
  341. return pdfManager.ensureCatalog("javaScript");
  342. });
  343. handler.on("GetOutline", function wphSetupGetOutline(data) {
  344. return pdfManager.ensureCatalog("documentOutline");
  345. });
  346. handler.on("GetOptionalContentConfig", function (data) {
  347. return pdfManager.ensureCatalog("optionalContentConfig");
  348. });
  349. handler.on("GetPermissions", function (data) {
  350. return pdfManager.ensureCatalog("permissions");
  351. });
  352. handler.on("GetMetadata", function wphSetupGetMetadata(data) {
  353. return Promise.all([pdfManager.ensureDoc("documentInfo"), pdfManager.ensureCatalog("metadata")]);
  354. });
  355. handler.on("GetData", function wphSetupGetData(data) {
  356. pdfManager.requestLoadedStream();
  357. return pdfManager.onLoadedStream().then(function (stream) {
  358. return stream.bytes;
  359. });
  360. });
  361. handler.on("GetStats", function wphSetupGetStats(data) {
  362. return pdfManager.ensureXRef("stats");
  363. });
  364. handler.on("GetAnnotations", function ({
  365. pageIndex,
  366. intent
  367. }) {
  368. return pdfManager.getPage(pageIndex).then(function (page) {
  369. return page.getAnnotationsData(intent);
  370. });
  371. });
  372. handler.on("SaveDocument", function ({
  373. numPages,
  374. annotationStorage,
  375. filename
  376. }) {
  377. pdfManager.requestLoadedStream();
  378. const promises = [pdfManager.onLoadedStream()];
  379. const document = pdfManager.pdfDocument;
  380. for (let pageIndex = 0; pageIndex < numPages; pageIndex++) {
  381. promises.push(pdfManager.getPage(pageIndex).then(function (page) {
  382. const task = new WorkerTask(`Save: page ${pageIndex}`);
  383. return page.save(handler, task, annotationStorage);
  384. }));
  385. }
  386. return Promise.all(promises).then(([stream, ...refs]) => {
  387. let newRefs = [];
  388. for (const ref of refs) {
  389. newRefs = ref.filter(x => x !== null).reduce((a, b) => a.concat(b), newRefs);
  390. }
  391. if (newRefs.length === 0) {
  392. return stream.bytes;
  393. }
  394. const xref = document.xref;
  395. let newXrefInfo = Object.create(null);
  396. if (xref.trailer) {
  397. const _info = Object.create(null);
  398. const xrefInfo = xref.trailer.get("Info") || null;
  399. if (xrefInfo) {
  400. xrefInfo.forEach((key, value) => {
  401. if ((0, _util.isString)(key) && (0, _util.isString)(value)) {
  402. _info[key] = (0, _util.stringToPDFString)(value);
  403. }
  404. });
  405. }
  406. newXrefInfo = {
  407. rootRef: xref.trailer.getRaw("Root") || null,
  408. encrypt: xref.trailer.getRaw("Encrypt") || null,
  409. newRef: xref.getNewRef(),
  410. infoRef: xref.trailer.getRaw("Info") || null,
  411. info: _info,
  412. fileIds: xref.trailer.getRaw("ID") || null,
  413. startXRef: document.startXRef,
  414. filename
  415. };
  416. }
  417. xref.resetNewRef();
  418. return (0, _writer.incrementalUpdate)(stream.bytes, newXrefInfo, newRefs);
  419. });
  420. });
  421. handler.on("GetOperatorList", function wphSetupRenderPage(data, sink) {
  422. var pageIndex = data.pageIndex;
  423. pdfManager.getPage(pageIndex).then(function (page) {
  424. var task = new WorkerTask(`GetOperatorList: page ${pageIndex}`);
  425. startWorkerTask(task);
  426. const start = verbosity >= _util.VerbosityLevel.INFOS ? Date.now() : 0;
  427. page.getOperatorList({
  428. handler,
  429. sink,
  430. task,
  431. intent: data.intent,
  432. renderInteractiveForms: data.renderInteractiveForms,
  433. annotationStorage: data.annotationStorage
  434. }).then(function (operatorListInfo) {
  435. finishWorkerTask(task);
  436. if (start) {
  437. (0, _util.info)(`page=${pageIndex + 1} - getOperatorList: time=` + `${Date.now() - start}ms, len=${operatorListInfo.length}`);
  438. }
  439. sink.close();
  440. }, function (reason) {
  441. finishWorkerTask(task);
  442. if (task.terminated) {
  443. return;
  444. }
  445. handler.send("UnsupportedFeature", {
  446. featureId: _util.UNSUPPORTED_FEATURES.errorOperatorList
  447. });
  448. sink.error(reason);
  449. });
  450. });
  451. }, this);
  452. handler.on("GetTextContent", function wphExtractText(data, sink) {
  453. var pageIndex = data.pageIndex;
  454. sink.onPull = function (desiredSize) {};
  455. sink.onCancel = function (reason) {};
  456. pdfManager.getPage(pageIndex).then(function (page) {
  457. var task = new WorkerTask("GetTextContent: page " + pageIndex);
  458. startWorkerTask(task);
  459. const start = verbosity >= _util.VerbosityLevel.INFOS ? Date.now() : 0;
  460. page.extractTextContent({
  461. handler,
  462. task,
  463. sink,
  464. normalizeWhitespace: data.normalizeWhitespace,
  465. combineTextItems: data.combineTextItems
  466. }).then(function () {
  467. finishWorkerTask(task);
  468. if (start) {
  469. (0, _util.info)(`page=${pageIndex + 1} - getTextContent: time=` + `${Date.now() - start}ms`);
  470. }
  471. sink.close();
  472. }, function (reason) {
  473. finishWorkerTask(task);
  474. if (task.terminated) {
  475. return;
  476. }
  477. sink.error(reason);
  478. });
  479. });
  480. });
  481. handler.on("FontFallback", function (data) {
  482. return pdfManager.fontFallback(data.id, handler);
  483. });
  484. handler.on("Cleanup", function wphCleanup(data) {
  485. return pdfManager.cleanup(true);
  486. });
  487. handler.on("Terminate", function wphTerminate(data) {
  488. terminated = true;
  489. const waitOn = [];
  490. if (pdfManager) {
  491. pdfManager.terminate(new _util.AbortException("Worker was terminated."));
  492. const cleanupPromise = pdfManager.cleanup();
  493. waitOn.push(cleanupPromise);
  494. pdfManager = null;
  495. } else {
  496. (0, _primitives.clearPrimitiveCaches)();
  497. }
  498. if (cancelXHRs) {
  499. cancelXHRs(new _util.AbortException("Worker was terminated."));
  500. }
  501. WorkerTasks.forEach(function (task) {
  502. waitOn.push(task.finished);
  503. task.terminate();
  504. });
  505. return Promise.all(waitOn).then(function () {
  506. handler.destroy();
  507. handler = null;
  508. });
  509. });
  510. handler.on("Ready", function wphReady(data) {
  511. setupDoc(docParams);
  512. docParams = null;
  513. });
  514. return workerHandlerName;
  515. }
  516. static initializeFromPort(port) {
  517. var handler = new _message_handler.MessageHandler("worker", "main", port);
  518. WorkerMessageHandler.setup(handler, port);
  519. handler.send("ready", null);
  520. }
  521. }
  522. exports.WorkerMessageHandler = WorkerMessageHandler;
  523. function isMessagePort(maybePort) {
  524. return typeof maybePort.postMessage === "function" && "onmessage" in maybePort;
  525. }
  526. if (typeof window === "undefined" && !_is_node.isNodeJS && typeof self !== "undefined" && isMessagePort(self)) {
  527. WorkerMessageHandler.initializeFromPort(self);
  528. }