UNPKG

22.8 kBJavaScriptView Raw
1/**
2 * @licstart The following is the entire license notice for the
3 * JavaScript code in this page
4 *
5 * Copyright 2022 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
24Object.defineProperty(exports, "__esModule", {
25 value: true
26});
27exports.WorkerTask = exports.WorkerMessageHandler = void 0;
28
29var _util = require("../shared/util.js");
30
31var _primitives = require("./primitives.js");
32
33var _core_utils = require("./core_utils.js");
34
35var _pdf_manager = require("./pdf_manager.js");
36
37var _cleanup_helper = require("./cleanup_helper.js");
38
39var _writer = require("./writer.js");
40
41var _is_node = require("../shared/is_node.js");
42
43var _message_handler = require("../shared/message_handler.js");
44
45var _worker_stream = require("./worker_stream.js");
46
47class WorkerTask {
48 constructor(name) {
49 this.name = name;
50 this.terminated = false;
51 this._capability = (0, _util.createPromiseCapability)();
52 }
53
54 get finished() {
55 return this._capability.promise;
56 }
57
58 finish() {
59 this._capability.resolve();
60 }
61
62 terminate() {
63 this.terminated = true;
64 }
65
66 ensureNotTerminated() {
67 if (this.terminated) {
68 throw new Error("Worker task was terminated");
69 }
70 }
71
72}
73
74exports.WorkerTask = WorkerTask;
75
76class WorkerMessageHandler {
77 static setup(handler, port) {
78 let testMessageProcessed = false;
79 handler.on("test", function wphSetupTest(data) {
80 if (testMessageProcessed) {
81 return;
82 }
83
84 testMessageProcessed = true;
85 handler.send("test", data instanceof Uint8Array);
86 });
87 handler.on("configure", function wphConfigure(data) {
88 (0, _util.setVerbosityLevel)(data.verbosity);
89 });
90 handler.on("GetDocRequest", function wphSetupDoc(data) {
91 return WorkerMessageHandler.createDocumentHandler(data, port);
92 });
93 }
94
95 static createDocumentHandler(docParams, port) {
96 let pdfManager;
97 let terminated = false;
98 let cancelXHRs = null;
99 const WorkerTasks = [];
100 const verbosity = (0, _util.getVerbosityLevel)();
101 const apiVersion = docParams.apiVersion;
102 const workerVersion = '2.15.349';
103
104 if (apiVersion !== workerVersion) {
105 throw new Error(`The API version "${apiVersion}" does not match ` + `the Worker version "${workerVersion}".`);
106 }
107
108 const enumerableProperties = [];
109
110 for (const property in []) {
111 enumerableProperties.push(property);
112 }
113
114 if (enumerableProperties.length) {
115 throw new Error("The `Array.prototype` contains unexpected enumerable properties: " + enumerableProperties.join(", ") + "; thus breaking e.g. `for...in` iteration of `Array`s.");
116 }
117
118 if (typeof ReadableStream === "undefined") {
119 const partialMsg = "The browser/environment lacks native support for critical " + "functionality used by the PDF.js library (e.g. `ReadableStream`); ";
120
121 if (_is_node.isNodeJS) {
122 throw new Error(partialMsg + "please use a `legacy`-build instead.");
123 }
124
125 throw new Error(partialMsg + "please update to a supported browser.");
126 }
127
128 const docId = docParams.docId;
129 const docBaseUrl = docParams.docBaseUrl;
130 const workerHandlerName = docParams.docId + "_worker";
131 let handler = new _message_handler.MessageHandler(workerHandlerName, docId, port);
132
133 function ensureNotTerminated() {
134 if (terminated) {
135 throw new Error("Worker was terminated");
136 }
137 }
138
139 function startWorkerTask(task) {
140 WorkerTasks.push(task);
141 }
142
143 function finishWorkerTask(task) {
144 task.finish();
145 const i = WorkerTasks.indexOf(task);
146 WorkerTasks.splice(i, 1);
147 }
148
149 async function loadDocument(recoveryMode) {
150 await pdfManager.ensureDoc("checkHeader");
151 await pdfManager.ensureDoc("parseStartXRef");
152 await pdfManager.ensureDoc("parse", [recoveryMode]);
153 await pdfManager.ensureDoc("checkFirstPage", [recoveryMode]);
154 await pdfManager.ensureDoc("checkLastPage", [recoveryMode]);
155 const isPureXfa = await pdfManager.ensureDoc("isPureXfa");
156
157 if (isPureXfa) {
158 const task = new WorkerTask("loadXfaFonts");
159 startWorkerTask(task);
160 await Promise.all([pdfManager.loadXfaFonts(handler, task).catch(reason => {}).then(() => finishWorkerTask(task)), pdfManager.loadXfaImages()]);
161 }
162
163 const [numPages, fingerprints] = await Promise.all([pdfManager.ensureDoc("numPages"), pdfManager.ensureDoc("fingerprints")]);
164 const htmlForXfa = isPureXfa ? await pdfManager.ensureDoc("htmlForXfa") : null;
165 return {
166 numPages,
167 fingerprints,
168 htmlForXfa
169 };
170 }
171
172 function getPdfManager(data, evaluatorOptions, enableXfa) {
173 const pdfManagerCapability = (0, _util.createPromiseCapability)();
174 let newPdfManager;
175 const source = data.source;
176
177 if (source.data) {
178 try {
179 newPdfManager = new _pdf_manager.LocalPdfManager(docId, source.data, source.password, handler, evaluatorOptions, enableXfa, docBaseUrl);
180 pdfManagerCapability.resolve(newPdfManager);
181 } catch (ex) {
182 pdfManagerCapability.reject(ex);
183 }
184
185 return pdfManagerCapability.promise;
186 }
187
188 let pdfStream,
189 cachedChunks = [];
190
191 try {
192 pdfStream = new _worker_stream.PDFWorkerStream(handler);
193 } catch (ex) {
194 pdfManagerCapability.reject(ex);
195 return pdfManagerCapability.promise;
196 }
197
198 const fullRequest = pdfStream.getFullReader();
199 fullRequest.headersReady.then(function () {
200 if (!fullRequest.isRangeSupported) {
201 return;
202 }
203
204 const disableAutoFetch = source.disableAutoFetch || fullRequest.isStreamingSupported;
205 newPdfManager = new _pdf_manager.NetworkPdfManager(docId, pdfStream, {
206 msgHandler: handler,
207 password: source.password,
208 length: fullRequest.contentLength,
209 disableAutoFetch,
210 rangeChunkSize: source.rangeChunkSize
211 }, evaluatorOptions, enableXfa, docBaseUrl);
212
213 for (const chunk of cachedChunks) {
214 newPdfManager.sendProgressiveData(chunk);
215 }
216
217 cachedChunks = [];
218 pdfManagerCapability.resolve(newPdfManager);
219 cancelXHRs = null;
220 }).catch(function (reason) {
221 pdfManagerCapability.reject(reason);
222 cancelXHRs = null;
223 });
224 let loaded = 0;
225
226 const flushChunks = function () {
227 const pdfFile = (0, _util.arraysToBytes)(cachedChunks);
228
229 if (source.length && pdfFile.length !== source.length) {
230 (0, _util.warn)("reported HTTP length is different from actual");
231 }
232
233 try {
234 newPdfManager = new _pdf_manager.LocalPdfManager(docId, pdfFile, source.password, handler, evaluatorOptions, enableXfa, docBaseUrl);
235 pdfManagerCapability.resolve(newPdfManager);
236 } catch (ex) {
237 pdfManagerCapability.reject(ex);
238 }
239
240 cachedChunks = [];
241 };
242
243 const readPromise = new Promise(function (resolve, reject) {
244 const readChunk = function ({
245 value,
246 done
247 }) {
248 try {
249 ensureNotTerminated();
250
251 if (done) {
252 if (!newPdfManager) {
253 flushChunks();
254 }
255
256 cancelXHRs = null;
257 return;
258 }
259
260 loaded += (0, _util.arrayByteLength)(value);
261
262 if (!fullRequest.isStreamingSupported) {
263 handler.send("DocProgress", {
264 loaded,
265 total: Math.max(loaded, fullRequest.contentLength || 0)
266 });
267 }
268
269 if (newPdfManager) {
270 newPdfManager.sendProgressiveData(value);
271 } else {
272 cachedChunks.push(value);
273 }
274
275 fullRequest.read().then(readChunk, reject);
276 } catch (e) {
277 reject(e);
278 }
279 };
280
281 fullRequest.read().then(readChunk, reject);
282 });
283 readPromise.catch(function (e) {
284 pdfManagerCapability.reject(e);
285 cancelXHRs = null;
286 });
287
288 cancelXHRs = function (reason) {
289 pdfStream.cancelAllRequests(reason);
290 };
291
292 return pdfManagerCapability.promise;
293 }
294
295 function setupDoc(data) {
296 function onSuccess(doc) {
297 ensureNotTerminated();
298 handler.send("GetDoc", {
299 pdfInfo: doc
300 });
301 }
302
303 function onFailure(ex) {
304 ensureNotTerminated();
305
306 if (ex instanceof _util.PasswordException) {
307 const task = new WorkerTask(`PasswordException: response ${ex.code}`);
308 startWorkerTask(task);
309 handler.sendWithPromise("PasswordRequest", ex).then(function ({
310 password
311 }) {
312 finishWorkerTask(task);
313 pdfManager.updatePassword(password);
314 pdfManagerReady();
315 }).catch(function () {
316 finishWorkerTask(task);
317 handler.send("DocException", ex);
318 });
319 } else if (ex instanceof _util.InvalidPDFException || ex instanceof _util.MissingPDFException || ex instanceof _util.UnexpectedResponseException || ex instanceof _util.UnknownErrorException) {
320 handler.send("DocException", ex);
321 } else {
322 handler.send("DocException", new _util.UnknownErrorException(ex.message, ex.toString()));
323 }
324 }
325
326 function pdfManagerReady() {
327 ensureNotTerminated();
328 loadDocument(false).then(onSuccess, function (reason) {
329 ensureNotTerminated();
330
331 if (!(reason instanceof _core_utils.XRefParseException)) {
332 onFailure(reason);
333 return;
334 }
335
336 pdfManager.requestLoadedStream();
337 pdfManager.onLoadedStream().then(function () {
338 ensureNotTerminated();
339 loadDocument(true).then(onSuccess, onFailure);
340 });
341 });
342 }
343
344 ensureNotTerminated();
345 const evaluatorOptions = {
346 maxImageSize: data.maxImageSize,
347 disableFontFace: data.disableFontFace,
348 ignoreErrors: data.ignoreErrors,
349 isEvalSupported: data.isEvalSupported,
350 fontExtraProperties: data.fontExtraProperties,
351 useSystemFonts: data.useSystemFonts,
352 cMapUrl: data.cMapUrl,
353 standardFontDataUrl: data.standardFontDataUrl
354 };
355 getPdfManager(data, evaluatorOptions, data.enableXfa).then(function (newPdfManager) {
356 if (terminated) {
357 newPdfManager.terminate(new _util.AbortException("Worker was terminated."));
358 throw new Error("Worker was terminated");
359 }
360
361 pdfManager = newPdfManager;
362 pdfManager.onLoadedStream().then(function (stream) {
363 handler.send("DataLoaded", {
364 length: stream.bytes.byteLength
365 });
366 });
367 }).then(pdfManagerReady, onFailure);
368 }
369
370 handler.on("GetPage", function wphSetupGetPage(data) {
371 return pdfManager.getPage(data.pageIndex).then(function (page) {
372 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]) {
373 return {
374 rotate,
375 ref,
376 userUnit,
377 view
378 };
379 });
380 });
381 });
382 handler.on("GetPageIndex", function wphSetupGetPageIndex(data) {
383 const pageRef = _primitives.Ref.get(data.num, data.gen);
384
385 return pdfManager.ensureCatalog("getPageIndex", [pageRef]);
386 });
387 handler.on("GetDestinations", function wphSetupGetDestinations(data) {
388 return pdfManager.ensureCatalog("destinations");
389 });
390 handler.on("GetDestination", function wphSetupGetDestination(data) {
391 return pdfManager.ensureCatalog("getDestination", [data.id]);
392 });
393 handler.on("GetPageLabels", function wphSetupGetPageLabels(data) {
394 return pdfManager.ensureCatalog("pageLabels");
395 });
396 handler.on("GetPageLayout", function wphSetupGetPageLayout(data) {
397 return pdfManager.ensureCatalog("pageLayout");
398 });
399 handler.on("GetPageMode", function wphSetupGetPageMode(data) {
400 return pdfManager.ensureCatalog("pageMode");
401 });
402 handler.on("GetViewerPreferences", function (data) {
403 return pdfManager.ensureCatalog("viewerPreferences");
404 });
405 handler.on("GetOpenAction", function (data) {
406 return pdfManager.ensureCatalog("openAction");
407 });
408 handler.on("GetAttachments", function wphSetupGetAttachments(data) {
409 return pdfManager.ensureCatalog("attachments");
410 });
411 handler.on("GetJavaScript", function wphSetupGetJavaScript(data) {
412 return pdfManager.ensureCatalog("javaScript");
413 });
414 handler.on("GetDocJSActions", function wphSetupGetDocJSActions(data) {
415 return pdfManager.ensureCatalog("jsActions");
416 });
417 handler.on("GetPageJSActions", function ({
418 pageIndex
419 }) {
420 return pdfManager.getPage(pageIndex).then(function (page) {
421 return pdfManager.ensure(page, "jsActions");
422 });
423 });
424 handler.on("GetOutline", function wphSetupGetOutline(data) {
425 return pdfManager.ensureCatalog("documentOutline");
426 });
427 handler.on("GetOptionalContentConfig", function (data) {
428 return pdfManager.ensureCatalog("optionalContentConfig");
429 });
430 handler.on("GetPermissions", function (data) {
431 return pdfManager.ensureCatalog("permissions");
432 });
433 handler.on("GetMetadata", function wphSetupGetMetadata(data) {
434 return Promise.all([pdfManager.ensureDoc("documentInfo"), pdfManager.ensureCatalog("metadata")]);
435 });
436 handler.on("GetMarkInfo", function wphSetupGetMarkInfo(data) {
437 return pdfManager.ensureCatalog("markInfo");
438 });
439 handler.on("GetData", function wphSetupGetData(data) {
440 pdfManager.requestLoadedStream();
441 return pdfManager.onLoadedStream().then(function (stream) {
442 return stream.bytes;
443 });
444 });
445 handler.on("GetAnnotations", function ({
446 pageIndex,
447 intent
448 }) {
449 return pdfManager.getPage(pageIndex).then(function (page) {
450 return page.getAnnotationsData(intent);
451 });
452 });
453 handler.on("GetFieldObjects", function (data) {
454 return pdfManager.ensureDoc("fieldObjects");
455 });
456 handler.on("HasJSActions", function (data) {
457 return pdfManager.ensureDoc("hasJSActions");
458 });
459 handler.on("GetCalculationOrderIds", function (data) {
460 return pdfManager.ensureDoc("calculationOrderIds");
461 });
462 handler.on("SaveDocument", function ({
463 isPureXfa,
464 numPages,
465 annotationStorage,
466 filename
467 }) {
468 pdfManager.requestLoadedStream();
469 const newAnnotationsByPage = !isPureXfa ? (0, _core_utils.getNewAnnotationsMap)(annotationStorage) : null;
470 const promises = [pdfManager.onLoadedStream(), pdfManager.ensureCatalog("acroForm"), pdfManager.ensureCatalog("acroFormRef"), pdfManager.ensureDoc("xref"), pdfManager.ensureDoc("startXRef")];
471
472 if (newAnnotationsByPage) {
473 for (const [pageIndex, annotations] of newAnnotationsByPage) {
474 promises.push(pdfManager.getPage(pageIndex).then(page => {
475 const task = new WorkerTask(`Save (editor): page ${pageIndex}`);
476 return page.saveNewAnnotations(handler, task, annotations).finally(function () {
477 finishWorkerTask(task);
478 });
479 }));
480 }
481 }
482
483 if (isPureXfa) {
484 promises.push(pdfManager.serializeXfaData(annotationStorage));
485 } else {
486 for (let pageIndex = 0; pageIndex < numPages; pageIndex++) {
487 promises.push(pdfManager.getPage(pageIndex).then(function (page) {
488 const task = new WorkerTask(`Save: page ${pageIndex}`);
489 return page.save(handler, task, annotationStorage).finally(function () {
490 finishWorkerTask(task);
491 });
492 }));
493 }
494 }
495
496 return Promise.all(promises).then(function ([stream, acroForm, acroFormRef, xref, startXRef, ...refs]) {
497 let newRefs = [];
498 let xfaData = null;
499
500 if (isPureXfa) {
501 xfaData = refs[0];
502
503 if (!xfaData) {
504 return stream.bytes;
505 }
506 } else {
507 newRefs = refs.flat(2);
508
509 if (newRefs.length === 0) {
510 return stream.bytes;
511 }
512 }
513
514 const xfa = acroForm instanceof _primitives.Dict && acroForm.get("XFA") || null;
515 let xfaDatasetsRef = null;
516 let hasXfaDatasetsEntry = false;
517
518 if (Array.isArray(xfa)) {
519 for (let i = 0, ii = xfa.length; i < ii; i += 2) {
520 if (xfa[i] === "datasets") {
521 xfaDatasetsRef = xfa[i + 1];
522 acroFormRef = null;
523 hasXfaDatasetsEntry = true;
524 }
525 }
526
527 if (xfaDatasetsRef === null) {
528 xfaDatasetsRef = xref.getNewRef();
529 }
530 } else if (xfa) {
531 acroFormRef = null;
532 (0, _util.warn)("Unsupported XFA type.");
533 }
534
535 let newXrefInfo = Object.create(null);
536
537 if (xref.trailer) {
538 const infoObj = Object.create(null);
539 const xrefInfo = xref.trailer.get("Info") || null;
540
541 if (xrefInfo instanceof _primitives.Dict) {
542 xrefInfo.forEach((key, value) => {
543 if (typeof value === "string") {
544 infoObj[key] = (0, _util.stringToPDFString)(value);
545 }
546 });
547 }
548
549 newXrefInfo = {
550 rootRef: xref.trailer.getRaw("Root") || null,
551 encryptRef: xref.trailer.getRaw("Encrypt") || null,
552 newRef: xref.getNewRef(),
553 infoRef: xref.trailer.getRaw("Info") || null,
554 info: infoObj,
555 fileIds: xref.trailer.get("ID") || null,
556 startXRef,
557 filename
558 };
559 }
560
561 xref.resetNewRef();
562 return (0, _writer.incrementalUpdate)({
563 originalData: stream.bytes,
564 xrefInfo: newXrefInfo,
565 newRefs,
566 xref,
567 hasXfa: !!xfa,
568 xfaDatasetsRef,
569 hasXfaDatasetsEntry,
570 acroFormRef,
571 acroForm,
572 xfaData
573 });
574 });
575 });
576 handler.on("GetOperatorList", function wphSetupRenderPage(data, sink) {
577 const pageIndex = data.pageIndex;
578 pdfManager.getPage(pageIndex).then(function (page) {
579 const task = new WorkerTask(`GetOperatorList: page ${pageIndex}`);
580 startWorkerTask(task);
581 const start = verbosity >= _util.VerbosityLevel.INFOS ? Date.now() : 0;
582 page.getOperatorList({
583 handler,
584 sink,
585 task,
586 intent: data.intent,
587 cacheKey: data.cacheKey,
588 annotationStorage: data.annotationStorage
589 }).then(function (operatorListInfo) {
590 finishWorkerTask(task);
591
592 if (start) {
593 (0, _util.info)(`page=${pageIndex + 1} - getOperatorList: time=` + `${Date.now() - start}ms, len=${operatorListInfo.length}`);
594 }
595
596 sink.close();
597 }, function (reason) {
598 finishWorkerTask(task);
599
600 if (task.terminated) {
601 return;
602 }
603
604 handler.send("UnsupportedFeature", {
605 featureId: _util.UNSUPPORTED_FEATURES.errorOperatorList
606 });
607 sink.error(reason);
608 });
609 });
610 });
611 handler.on("GetTextContent", function wphExtractText(data, sink) {
612 const pageIndex = data.pageIndex;
613 pdfManager.getPage(pageIndex).then(function (page) {
614 const task = new WorkerTask("GetTextContent: page " + pageIndex);
615 startWorkerTask(task);
616 const start = verbosity >= _util.VerbosityLevel.INFOS ? Date.now() : 0;
617 page.extractTextContent({
618 handler,
619 task,
620 sink,
621 includeMarkedContent: data.includeMarkedContent,
622 combineTextItems: data.combineTextItems
623 }).then(function () {
624 finishWorkerTask(task);
625
626 if (start) {
627 (0, _util.info)(`page=${pageIndex + 1} - getTextContent: time=` + `${Date.now() - start}ms`);
628 }
629
630 sink.close();
631 }, function (reason) {
632 finishWorkerTask(task);
633
634 if (task.terminated) {
635 return;
636 }
637
638 sink.error(reason);
639 });
640 });
641 });
642 handler.on("GetStructTree", function wphGetStructTree(data) {
643 return pdfManager.getPage(data.pageIndex).then(function (page) {
644 return pdfManager.ensure(page, "getStructTree");
645 });
646 });
647 handler.on("FontFallback", function (data) {
648 return pdfManager.fontFallback(data.id, handler);
649 });
650 handler.on("Cleanup", function wphCleanup(data) {
651 return pdfManager.cleanup(true);
652 });
653 handler.on("Terminate", function wphTerminate(data) {
654 terminated = true;
655 const waitOn = [];
656
657 if (pdfManager) {
658 pdfManager.terminate(new _util.AbortException("Worker was terminated."));
659 const cleanupPromise = pdfManager.cleanup();
660 waitOn.push(cleanupPromise);
661 pdfManager = null;
662 } else {
663 (0, _cleanup_helper.clearGlobalCaches)();
664 }
665
666 if (cancelXHRs) {
667 cancelXHRs(new _util.AbortException("Worker was terminated."));
668 }
669
670 for (const task of WorkerTasks) {
671 waitOn.push(task.finished);
672 task.terminate();
673 }
674
675 return Promise.all(waitOn).then(function () {
676 handler.destroy();
677 handler = null;
678 });
679 });
680 handler.on("Ready", function wphReady(data) {
681 setupDoc(docParams);
682 docParams = null;
683 });
684 return workerHandlerName;
685 }
686
687 static initializeFromPort(port) {
688 const handler = new _message_handler.MessageHandler("worker", "main", port);
689 WorkerMessageHandler.setup(handler, port);
690 handler.send("ready", null);
691 }
692
693}
694
695exports.WorkerMessageHandler = WorkerMessageHandler;
696
697function isMessagePort(maybePort) {
698 return typeof maybePort.postMessage === "function" && "onmessage" in maybePort;
699}
700
701if (typeof window === "undefined" && !_is_node.isNodeJS && typeof self !== "undefined" && isMessagePort(self)) {
702 WorkerMessageHandler.initializeFromPort(self);
703}
\No newline at end of file