Merge branch 'develop' into devsecops

aleks-ivanov · May 7, 2024 · 978525e · 978525e
2 parents 402f78a + 7755191
commit 978525e
Show file tree

Hide file tree

Showing 9 changed files with 119 additions and 15 deletions.
diff --git a/itext.tests/itext.kernel.tests/itext/kernel/pdf/ParentTreeTest.cs b/itext.tests/itext.kernel.tests/itext/kernel/pdf/ParentTreeTest.cs
@@ -333,6 +333,85 @@ public virtual void XObjDoesntHaveStructParentTest() {
                 ));
         }
 
+        [NUnit.Framework.Test]
+        [LogMessage(iText.IO.Logs.IoLogMessageConstant.TAG_STRUCTURE_INIT_FAILED)]
+        public virtual void ObjRefNoStructParentNoModificationTest() {
+            String pdf = sourceFolder + "objRefNoStructParent.pdf";
+            String outPdf = destinationFolder + "objRefNoStructParentNoModification.pdf";
+            PdfReader reader = new PdfReader(pdf).SetStrictnessLevel(PdfReader.StrictnessLevel.CONSERVATIVE);
+            PdfDocument doc = new PdfDocument(reader, CompareTool.CreateTestPdfWriter(outPdf));
+            PdfArray nums = doc.GetCatalog().GetPdfObject().GetAsDictionary(PdfName.StructTreeRoot).GetAsDictionary(PdfName
+                .ParentTree).GetAsArray(PdfName.Nums);
+            NUnit.Framework.Assert.IsNull(GetStructParentEntry(nums.Get(3)));
+            NUnit.Framework.Assert.IsNull(GetStructParentEntry(nums.Get(5)));
+            NUnit.Framework.Assert.IsNull(GetStructParentEntry(nums.Get(7)));
+            NUnit.Framework.Assert.IsNull(GetStructParentEntry(nums.Get(9)));
+            doc.Close();
+        }
+
+        [NUnit.Framework.Test]
+        [LogMessage(KernelLogMessageConstant.STRUCT_PARENT_INDEX_MISSED_AND_RECREATED, Count = 4)]
+        public virtual void ObjRefNoStructParentModificationTest() {
+            String pdf = sourceFolder + "objRefNoStructParent.pdf";
+            String outPdf = destinationFolder + "objRefNoStructParentModification.pdf";
+            String cmpPdf = sourceFolder + "cmp_objRefNoStructParentModification.pdf";
+            PdfDocument doc = new PdfDocument(new PdfReader(pdf), CompareTool.CreateTestPdfWriter(outPdf));
+            doc.Close();
+            NUnit.Framework.Assert.IsNull(new CompareTool().CompareByContent(outPdf, cmpPdf, destinationFolder, "diff"
+                ));
+        }
+
+        [NUnit.Framework.Test]
+        [LogMessage(iText.IO.Logs.IoLogMessageConstant.TAG_STRUCTURE_INIT_FAILED)]
+        public virtual void XObjNoStructParentNoModificationTest() {
+            String pdf = sourceFolder + "xObjNoStructParent.pdf";
+            String outPdf = destinationFolder + "xObjNoStructParentNoModification.pdf";
+            PdfReader reader = new PdfReader(pdf).SetStrictnessLevel(PdfReader.StrictnessLevel.CONSERVATIVE);
+            PdfDocument doc = new PdfDocument(reader, new PdfWriter(outPdf));
+            PdfObject obj = doc.GetCatalog().GetPdfObject().GetAsDictionary(PdfName.StructTreeRoot).GetAsDictionary(PdfName
+                .ParentTree).GetAsArray(PdfName.Nums).Get(1);
+            PdfStream xObj = ((PdfDictionary)((PdfArray)obj).Get(0)).GetAsDictionary(PdfName.K).GetAsStream(PdfName.Stm
+                );
+            NUnit.Framework.Assert.IsNull(xObj.Get(PdfName.StructParent));
+            doc.Close();
+        }
+
+        [NUnit.Framework.Test]
+        [LogMessage(KernelLogMessageConstant.XOBJECT_STRUCT_PARENT_INDEX_MISSED_AND_RECREATED)]
+        public virtual void XObjNoStructParentModificationTest() {
+            String pdf = sourceFolder + "xObjNoStructParent.pdf";
+            String outPdf = destinationFolder + "xObjNoStructParentModification.pdf";
+            String cmpPdf = sourceFolder + "cmp_xObjNoStructParentModification.pdf";
+            PdfDocument doc = new PdfDocument(new PdfReader(pdf), CompareTool.CreateTestPdfWriter(outPdf));
+            doc.Close();
+            NUnit.Framework.Assert.IsNull(new CompareTool().CompareByContent(outPdf, cmpPdf, destinationFolder, "diff"
+                ));
+        }
+
+        [NUnit.Framework.Test]
+        [LogMessage(KernelLogMessageConstant.STRUCT_PARENT_INDEX_MISSED_AND_RECREATED)]
+        public virtual void ObjRefNoStructParentNoReaderTest() {
+            String outPdf = destinationFolder + "objRefNoStructParentNoReader.pdf";
+            String cmpPdf = sourceFolder + "cmp_objRefNoStructParentNoReader.pdf";
+            PdfDocument pdfDoc = new PdfDocument(CompareTool.CreateTestPdfWriter(outPdf));
+            pdfDoc.SetTagged();
+            PdfPage page = pdfDoc.AddNewPage();
+            PdfDictionary mcrDic = new PdfDictionary();
+            mcrDic.Put(PdfName.Pg, page.GetPdfObject());
+            mcrDic.Put(PdfName.MCID, new PdfNumber(0));
+            mcrDic.Put(PdfName.Obj, new PdfDictionary());
+            PdfDictionary elemDic = new PdfDictionary();
+            elemDic.Put(PdfName.P, pdfDoc.GetStructTreeRoot().GetPdfObject());
+            PdfStructElem elem = new PdfStructElem(elemDic);
+            elem.MakeIndirect(pdfDoc);
+            PdfMcr mcr = new PdfObjRef(mcrDic, elem);
+            elem.AddKid(0, mcr);
+            pdfDoc.GetStructTreeRoot().AddKid(elem);
+            pdfDoc.Close();
+            NUnit.Framework.Assert.IsNull(new CompareTool().CompareByContent(outPdf, cmpPdf, destinationFolder, "diff"
+                ));
+        }
+
         [NUnit.Framework.Test]
         [LogMessage(iText.IO.Logs.IoLogMessageConstant.CREATED_ROOT_TAG_HAS_MAPPING)]
         public virtual void CopyPageWithMultipleDocumentTagsTest() {
@@ -341,6 +420,11 @@ public virtual void CopyPageWithMultipleDocumentTagsTest() {
             NUnit.Framework.Assert.DoesNotThrow(() => pdfDoc.GetTagStructureContext().NormalizeDocumentRootTag());
         }
 
+        private PdfObject GetStructParentEntry(PdfObject obj) {
+            return ((PdfDictionary)obj).GetAsDictionary(PdfName.K).GetAsDictionary(PdfName.Obj).Get(PdfName.StructParent
+                );
+        }
+
         private bool CheckParentTree(String outFileName, String cmpFileName) {
             PdfReader outReader = CompareTool.CreateOutputReader(outFileName);
             PdfDocument outDocument = new PdfDocument(outReader);

diff --git a/....tests/resources/itext/kernel/pdf/ParentTreeTest/cmp_objRefNoStructParentModification.pdf b/....tests/resources/itext/kernel/pdf/ParentTreeTest/cmp_objRefNoStructParentModification.pdf
diff --git a/...rnel.tests/resources/itext/kernel/pdf/ParentTreeTest/cmp_objRefNoStructParentNoReader.pdf b/...rnel.tests/resources/itext/kernel/pdf/ParentTreeTest/cmp_objRefNoStructParentNoReader.pdf
diff --git a/...el.tests/resources/itext/kernel/pdf/ParentTreeTest/cmp_xObjNoStructParentModification.pdf b/...el.tests/resources/itext/kernel/pdf/ParentTreeTest/cmp_xObjNoStructParentModification.pdf
diff --git a/...sts/itext.kernel.tests/resources/itext/kernel/pdf/ParentTreeTest/objRefNoStructParent.pdf b/...sts/itext.kernel.tests/resources/itext/kernel/pdf/ParentTreeTest/objRefNoStructParent.pdf
diff --git a/...tests/itext.kernel.tests/resources/itext/kernel/pdf/ParentTreeTest/xObjNoStructParent.pdf b/...tests/itext.kernel.tests/resources/itext/kernel/pdf/ParentTreeTest/xObjNoStructParent.pdf
diff --git a/itext/itext.kernel/itext/kernel/exceptions/KernelExceptionMessageConstant.cs b/itext/itext.kernel/itext/kernel/exceptions/KernelExceptionMessageConstant.cs
@@ -445,10 +445,8 @@ public const String CONTENT_STREAM_MUST_NOT_INVOKE_OPERATORS_THAT_SPECIFY_COLORS
 
         public const String STREAM_SHALL_END_WITH_ENDSTREAM = "Stream shall end with endstream keyword.";
 
-        [Obsolete]
-        public const String STRUCT_PARENT_INDEX_NOT_FOUND_IN_TAGGED_OBJECT = 
-                // Replaced with log message
-                "StructParent index not found in " + "tagged object.";
+        public const String STRUCT_PARENT_INDEX_NOT_FOUND_IN_TAGGED_OBJECT = "StructParent index not found in " + 
+            "tagged object.";
 
         public const String STRUCTURE_ELEMENT_IN_STRUCTURE_DESTINATION_SHALL_BE_AN_INDIRECT_OBJECT = "Structure " 
             + "element referenced by a structure destination shall be an indirect object.";
@@ -544,6 +542,8 @@ public const String WHEN_ADDING_OBJECT_REFERENCE_TO_THE_TAG_TREE_IT_MUST_BE_CONN
         public const String XREF_STRUCTURE_SIZE_EXCEEDED_THE_LIMIT = "Xref structure contains too many elements " 
             + "and may cause OOM exception. You can increase number of elements by setting custom " + "MemoryLimitsAwareHandler.";
 
+        public const String XOBJECT_STRUCT_PARENT_INDEX_MISSED = "XObject has no StructParents index in its stream.";
+
         public const String TOTAL_XOBJECT_SIZE_ONE_PAGE_EXCEEDED_THE_LIMIT = "Pdf contains too many xObject elements on a page "
              + "and may cause OOM exception. You can increase page size limit by setting custom " + "MemoryLimitsAwareHandler.";
 

diff --git a/itext/itext.kernel/itext/kernel/pdf/tagging/ParentTreeHandler.cs b/itext/itext.kernel/itext/kernel/pdf/tagging/ParentTreeHandler.cs
@@ -165,11 +165,16 @@ private void RegisterMcr(PdfMcr mcr, bool registeringOnInit) {
                     }
                 }
                 else {
-                    maxStructParentIndex++;
-                    xObjectToStructParentsInd.Put(stmIndRef, maxStructParentIndex);
-                    xObjectStream.Put(PdfName.StructParents, new PdfNumber(maxStructParentIndex));
-                    structTreeRoot.GetPdfObject().Put(PdfName.ParentTreeNextKey, new PdfNumber(maxStructParentIndex + 1));
-                    LOGGER.LogWarning(KernelLogMessageConstant.XOBJECT_STRUCT_PARENT_INDEX_MISSED_AND_RECREATED);
+                    if (IsModificationAllowed()) {
+                        maxStructParentIndex++;
+                        xObjectToStructParentsInd.Put(stmIndRef, maxStructParentIndex);
+                        xObjectStream.Put(PdfName.StructParents, new PdfNumber(maxStructParentIndex));
+                        structTreeRoot.GetPdfObject().Put(PdfName.ParentTreeNextKey, new PdfNumber(maxStructParentIndex + 1));
+                        LOGGER.LogWarning(KernelLogMessageConstant.XOBJECT_STRUCT_PARENT_INDEX_MISSED_AND_RECREATED);
+                    }
+                    else {
+                        throw new PdfException(KernelExceptionMessageConstant.XOBJECT_STRUCT_PARENT_INDEX_MISSED);
+                    }
                 }
                 pageMcrs.PutXObjectMcr(stmIndRef, mcr);
             }
@@ -185,11 +190,16 @@ private void RegisterMcr(PdfMcr mcr, bool registeringOnInit) {
                         pageMcrs.PutObjectReferenceMcr(n.IntValue(), mcr);
                     }
                     else {
-                        maxStructParentIndex++;
-                        pageMcrs.PutObjectReferenceMcr(maxStructParentIndex, mcr);
-                        obj.Put(PdfName.StructParent, new PdfNumber(maxStructParentIndex));
-                        structTreeRoot.GetPdfObject().Put(PdfName.ParentTreeNextKey, new PdfNumber(maxStructParentIndex + 1));
-                        LOGGER.LogWarning(KernelLogMessageConstant.STRUCT_PARENT_INDEX_MISSED_AND_RECREATED);
+                        if (IsModificationAllowed()) {
+                            maxStructParentIndex++;
+                            pageMcrs.PutObjectReferenceMcr(maxStructParentIndex, mcr);
+                            obj.Put(PdfName.StructParent, new PdfNumber(maxStructParentIndex));
+                            structTreeRoot.GetPdfObject().Put(PdfName.ParentTreeNextKey, new PdfNumber(maxStructParentIndex + 1));
+                            LOGGER.LogWarning(KernelLogMessageConstant.STRUCT_PARENT_INDEX_MISSED_AND_RECREATED);
+                        }
+                        else {
+                            throw new PdfException(KernelExceptionMessageConstant.STRUCT_PARENT_INDEX_NOT_FOUND_IN_TAGGED_OBJECT);
+                        }
                     }
                 }
                 else {
@@ -251,6 +261,16 @@ public virtual void UnregisterMcr(PdfMcr mcrToUnregister) {
             }
         }
 
+        private bool IsModificationAllowed() {
+            PdfReader reader = this.structTreeRoot.GetDocument().GetReader();
+            if (reader != null) {
+                return PdfReader.StrictnessLevel.CONSERVATIVE.IsStricter(reader.GetStrictnessLevel());
+            }
+            else {
+                return true;
+            }
+        }
+
         private void RegisterAllMcrs() {
             pageToPageMcrs = new Dictionary<PdfIndirectReference, ParentTreeHandler.PageMcrsContainer>();
             // we create new number tree and not using parentTree, because we want parentTree to be empty

diff --git a/port-hash b/port-hash
@@ -1 +1 @@
-66f84b927f433f9f4d9d5ff5bda6a3b6d2dec5d4
+9ad717077d0cc1a30ebd87cdd84bfecc36cf0ac4
Original file line number	Diff line number	Diff line change
		@@ -1 +1 @@
		66f84b927f433f9f4d9d5ff5bda6a3b6d2dec5d4
		9ad717077d0cc1a30ebd87cdd84bfecc36cf0ac4