Allow getOperatorList/getTextContent to skip errors when parsing broken XObjects (issue 8702, issue 8704)

This patch makes use of the existing `ignoreErrors` property in `src/core/evaluator.js`, see PRs 8240 and 8441, thus allowing us to attempt to recovery as much as possible of a page even when it contains broken XObjects.

Fixes 8702.
Fixes 8704.

src/core/evaluator.js

@@ -948,52 +948,65 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
             case OPS.paintXObject:
               // eagerly compile XForm objects
               var name = args[0].name;
-              if (!name) {
+              if (name && imageCache[name] !== undefined) {
-                warn('XObject must be referred to by name.');
-                continue;
-              }
-              if (imageCache[name] !== undefined) {
                 operatorList.addOp(imageCache[name].fn, imageCache[name].args);
                 args = null;
                 continue;
               }
 
-              var xobj = xobjs.get(name);
+              next(new Promise(function(resolveXObject, rejectXObject) {
-              if (xobj) {
+                if (!name) {
+                  throw new FormatError('XObject must be referred to by name.');
+                }
+
+                let xobj = xobjs.get(name);
+                if (!xobj) {
+                  operatorList.addOp(fn, args);
+                  resolveXObject();
+                  return;
+                }
                 if (!isStream(xobj)) {
                   throw new FormatError('XObject should be a stream');
                 }
 
-                var type = xobj.dict.get('Subtype');
+                let type = xobj.dict.get('Subtype');
                 if (!isName(type)) {
                   throw new FormatError('XObject should have a Name subtype');
                 }
 
                 if (type.name === 'Form') {
                   stateManager.save();
-                  next(self.buildFormXObject(resources, xobj, null,
+                  self.buildFormXObject(resources, xobj, null, operatorList,
-                                             operatorList, task,
+                                        task, stateManager.state.clone()).
-                                             stateManager.state.clone()).
+                    then(function() {
-                    then(function () {
                       stateManager.restore();
-                    }));
+                      resolveXObject();
+                    }, rejectXObject);
                   return;
                 } else if (type.name === 'Image') {
                   self.buildPaintImageXObject(resources, xobj, false,
                                               operatorList, name, imageCache);
-                  args = null;
-                  continue;
                 } else if (type.name === 'PS') {
                   // PostScript XObjects are unused when viewing documents.
                   // See section 4.7.1 of Adobe's PDF reference.
                   info('Ignored XObject subtype PS');
-                  continue;
                 } else {
                   throw new FormatError(
                     `Unhandled XObject subtype ${type.name}`);
                 }
+                resolveXObject();
+              }).catch(function(reason) {
+                if (self.options.ignoreErrors) {
+                  // Error(s) in the XObject -- sending unsupported feature
+                  // notification and allow rendering to continue.
+                  self.handler.send('UnsupportedFeature',
+                    { featureId: UNSUPPORTED_FEATURES.unknown, });
+                  warn(`getOperatorList - ignoring XObject: "${reason}".`);
+                  return;
                 }
-              break;
+                throw reason;
+              }));
+              return;
             case OPS.setFont:
               var fontSize = args[1];
               // eagerly collect all fonts
@@ -1666,36 +1679,43 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
               }
 
               var name = args[0].name;
-              if (name in skipEmptyXObjs) {
+              if (name && skipEmptyXObjs[name] !== undefined) {
                 break;
               }
 
-              var xobj = xobjs.get(name);
+              next(new Promise(function(resolveXObject, rejectXObject) {
+                if (!name) {
+                  throw new FormatError('XObject must be referred to by name.');
+                }
+
+                let xobj = xobjs.get(name);
                 if (!xobj) {
-                break;
+                  resolveXObject();
+                  return;
                 }
                 if (!isStream(xobj)) {
                   throw new FormatError('XObject should be a stream');
                 }
 
-              var type = xobj.dict.get('Subtype');
+                let type = xobj.dict.get('Subtype');
                 if (!isName(type)) {
                   throw new FormatError('XObject should have a Name subtype');
                 }
 
                 if (type.name !== 'Form') {
                   skipEmptyXObjs[name] = true;
-                break;
+                  resolveXObject();
+                  return;
                 }
 
                 // Use a new `StateManager` to prevent incorrect positioning of
                 // textItems *after* the Form XObject, since errors in the data
-              // can otherwise prevent `restore` operators from being executed.
+                // can otherwise prevent `restore` operators from executing.
-              // NOTE: This is only an issue when `options.ignoreErrors = true`.
+                // NOTE: Only an issue when `options.ignoreErrors === true`.
-              var currentState = stateManager.state.clone();
+                let currentState = stateManager.state.clone();
-              var xObjStateManager = new StateManager(currentState);
+                let xObjStateManager = new StateManager(currentState);
 
-              var matrix = xobj.dict.getArray('Matrix');
+                let matrix = xobj.dict.getArray('Matrix');
                 if (Array.isArray(matrix) && matrix.length === 6) {
                   xObjStateManager.transform(matrix);
                 }
@@ -1720,7 +1740,7 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
                   },
                 };
 
-              next(self.getTextContent({
+                self.getTextContent({
                   stream: xobj,
                   task,
                   resources: xobj.dict.get('Resources') || resources,
@@ -1733,6 +1753,19 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
                   if (!sinkWrapper.enqueueInvoked) {
                     skipEmptyXObjs[name] = true;
                   }
+                  resolveXObject();
+                }, rejectXObject);
+              }).catch(function(reason) {
+                if (reason instanceof AbortException) {
+                  return;
+                }
+                if (self.options.ignoreErrors) {
+                  // Error(s) in the XObject -- allow text-extraction to
+                  // continue.
+                  warn(`getTextContent - ignoring XObject: "${reason}".`);
+                  return;
+                }
+                throw reason;
               }));
               return;
             case OPS.setGState:

test/pdfs/.gitignore

@@ -57,6 +57,7 @@
 !issue8480.pdf
 !issue8570.pdf
 !issue8697.pdf
+!issue8702.pdf
 !issue8707.pdf
 !issue8798r.pdf
 !issue8823.pdf

test/test_manifest.json

@@ -1640,6 +1640,22 @@
        "lastPage": 1,
        "type": "load"
     },
+    {  "id": "issue8702-eq",
+       "file": "pdfs/issue8702.pdf",
+       "md5": "59d501ed1518d78ef6ee442cf824b0f6",
+       "rounds": 1,
+       "link": false,
+       "lastPage": 1,
+       "type": "eq"
+    },
+    {  "id": "issue8702-text",
+       "file": "pdfs/issue8702.pdf",
+       "md5": "59d501ed1518d78ef6ee442cf824b0f6",
+       "rounds": 1,
+       "link": false,
+       "lastPage": 1,
+       "type": "text"
+    },
     {  "id": "pr4897",
        "file": "pdfs/pr4897.pdf",
        "md5": "26897633eea5e6d10345a130b1c1777c",

test/unit/evaluator_spec.js

@@ -14,9 +14,9 @@
  */
 
 import { Dict, Name } from '../../src/core/primitives';
+import { FormatError, OPS } from '../../src/shared/util';
 import { OperatorList, PartialEvaluator } from '../../src/core/evaluator';
 import { Stream, StringStream } from '../../src/core/stream';
-import { OPS } from '../../src/shared/util';
 import { WorkerTask } from '../../src/core/worker';
 import { XRefMock } from './test_utils';
 
@@ -48,6 +48,8 @@ describe('evaluator', function() {
       operatorList: result,
     }).then(function() {
       callback(result);
+    }, function(reason) {
+      callback(reason);
     });
   }
 
@@ -229,9 +231,9 @@ describe('evaluator', function() {
     it('should skip paintXObject if name is missing', function(done) {
       var stream = new StringStream('/ Do');
       runOperatorListCheck(partialEvaluator, stream, new ResourcesMock(),
-          function (result) {
+          function(result) {
-        expect(result.argsArray).toEqual([]);
+        expect(result instanceof FormatError).toEqual(true);
-        expect(result.fnArray).toEqual([]);
+        expect(result.message).toEqual('XObject must be referred to by name.');
         done();
       });
     });