Support StructTreeRoot comparison, fix minor bugs

DEVSIX-8413

Autoported commit.
Original commit hash: [e0488e9f9]

Author: Eugene Bochilo

itext.tests/itext.io.tests/itext/io/source/PdfTokenizerTest.cs

@@ -195,6 +195,23 @@ public virtual void GetNextEofLongTextTest() {
             }
         }
 
+        [NUnit.Framework.Test]
+        public virtual void GetNextEofWhichIsCutTest() {
+            StringBuilder stringBuilder = new StringBuilder();
+            // We append 'a' 124 times because buffer has 128 bytes length.
+            // This way '%%EOF' is cut and first string only contains '%%EO'
+            for (int i = 0; i < 124; ++i) {
+                stringBuilder.Append("a");
+            }
+            stringBuilder.Append("%%EOF");
+            RandomAccessSourceFactory factory = new RandomAccessSourceFactory();
+            using (PdfTokenizer tok = new PdfTokenizer(new RandomAccessFileOrArray(factory.CreateSource(stringBuilder.
+                ToString().GetBytes(iText.Commons.Utils.EncodingUtil.ISO_8859_1))))) {
+                long eofPosition = tok.GetNextEof();
+                NUnit.Framework.Assert.AreEqual(124 + 6, eofPosition);
+            }
+        }
+
         [NUnit.Framework.Test]
         public virtual void GetNextEofSeveralEofTest() {
             String data = "some text %%EOFto test \nget%%EOFting end of\n fil%%EOFe logic%%EOF";
@@ -206,6 +223,16 @@ public virtual void GetNextEofSeveralEofTest() {
             }
         }
 
+        [NUnit.Framework.Test]
+        public virtual void GetNextEofNoEofTest() {
+            String data = "some text to test \ngetting end of\n file logic";
+            RandomAccessSourceFactory factory = new RandomAccessSourceFactory();
+            using (PdfTokenizer tok = new PdfTokenizer(new RandomAccessFileOrArray(factory.CreateSource(data.GetBytes(
+                iText.Commons.Utils.EncodingUtil.ISO_8859_1))))) {
+                NUnit.Framework.Assert.Catch(typeof(iText.IO.Exceptions.IOException), () => tok.GetNextEof());
+            }
+        }
+
         [NUnit.Framework.Test]
         public virtual void GetDecodedStringContentTest() {
             String data = "/Name1 15";

itext.tests/itext.kernel.tests/itext/kernel/pdf/PdfRevisionsReaderTest.cs

@@ -161,7 +161,7 @@ public virtual void DocumentWithStreamAndTableXref() {
                 NUnit.Framework.Assert.AreEqual(1381, thirdRevision.GetEofOffset());
                 DocumentRevision secondRevision = revisionsReader.GetAllRevisions()[1];
                 AssertResultingRevision(secondRevision, 1, 2, 3, 4, 5, 6, 7, 8);
-                NUnit.Framework.Assert.AreEqual(1550, secondRevision.GetEofOffset());
+                NUnit.Framework.Assert.AreEqual(1381, secondRevision.GetEofOffset());
                 DocumentRevision firstRevision = revisionsReader.GetAllRevisions()[2];
                 AssertResultingRevision(firstRevision);
                 NUnit.Framework.Assert.AreEqual(1550, firstRevision.GetEofOffset());

itext.tests/itext.sign.tests/itext/signatures/validation/v1/DocumentRevisionsValidatorIntegrationTest.cs

@@ -410,6 +410,105 @@ public virtual void RemoveAllTheFieldAnnotationsTest() {
             }
         }
 
+        [NUnit.Framework.Test]
+        public virtual void SimpleTaggedDocTest() {
+            using (PdfDocument document = new PdfDocument(new PdfReader(SOURCE_FOLDER + "simpleTaggedDoc.pdf"))) {
+                DocumentRevisionsValidator validator = builder.BuildDocumentRevisionsValidator();
+                ValidationReport report = validator.ValidateAllDocumentRevisions(validationContext, document);
+                AssertValidationReport.AssertThat(report, (a) => a.HasStatus(ValidationReport.ValidationResult.VALID).HasNumberOfLogs
+                    (0));
+            }
+        }
+
+        [NUnit.Framework.Test]
+        public virtual void TaggedDocAddAndRemoveAnnotationsTest() {
+            using (PdfDocument document = new PdfDocument(new PdfReader(SOURCE_FOLDER + "taggedDocAddAndRemoveAnnotations.pdf"
+                ))) {
+                DocumentRevisionsValidator validator = builder.BuildDocumentRevisionsValidator();
+                ValidationReport report = validator.ValidateAllDocumentRevisions(validationContext, document);
+                // Annotations were removed, but were also considered modified objects and therefore are added to xref table.
+                AssertValidationReport.AssertThat(report, (a) => a.HasStatus(ValidationReport.ValidationResult.VALID).HasNumberOfLogs
+                    (2).HasLogItem((l) => l.WithCheckName(DocumentRevisionsValidator.DOC_MDP_CHECK).WithMessage(DocumentRevisionsValidator
+                    .UNEXPECTED_ENTRY_IN_XREF, (m) => "18").WithStatus(ReportItem.ReportItemStatus.INFO)).HasLogItem((l) =>
+                     l.WithCheckName(DocumentRevisionsValidator.DOC_MDP_CHECK).WithMessage(DocumentRevisionsValidator.UNEXPECTED_ENTRY_IN_XREF
+                    , (m) => "50").WithStatus(ReportItem.ReportItemStatus.INFO)));
+            }
+        }
+
+        [NUnit.Framework.Test]
+        public virtual void TaggedDocRemoveStructTreeElementTest() {
+            using (PdfDocument document = new PdfDocument(new PdfReader(SOURCE_FOLDER + "taggedDocRemoveStructTreeElement.pdf"
+                ))) {
+                DocumentRevisionsValidator validator = builder.BuildDocumentRevisionsValidator();
+                ValidationReport report = validator.ValidateAllDocumentRevisions(validationContext, document);
+                AssertValidationReport.AssertThat(report, (a) => a.HasStatus(ValidationReport.ValidationResult.INVALID).HasNumberOfFailures
+                    (1).HasLogItem((l) => l.WithCheckName(DocumentRevisionsValidator.DOC_MDP_CHECK).WithMessage(DocumentRevisionsValidator
+                    .STRUCT_TREE_ROOT_MODIFIED).WithStatus(ReportItem.ReportItemStatus.INVALID)));
+            }
+        }
+
+        [NUnit.Framework.Test]
+        public virtual void TaggedDocRemoveStructTreeAnnotationTest() {
+            using (PdfDocument document = new PdfDocument(new PdfReader(SOURCE_FOLDER + "taggedDocRemoveStructTreeAnnotation.pdf"
+                ))) {
+                DocumentRevisionsValidator validator = builder.BuildDocumentRevisionsValidator();
+                ValidationReport report = validator.ValidateAllDocumentRevisions(validationContext, document);
+                AssertValidationReport.AssertThat(report, (a) => a.HasStatus(ValidationReport.ValidationResult.INVALID).HasNumberOfFailures
+                    (1).HasLogItem((l) => l.WithCheckName(DocumentRevisionsValidator.DOC_MDP_CHECK).WithMessage(DocumentRevisionsValidator
+                    .STRUCT_TREE_ROOT_MODIFIED).WithStatus(ReportItem.ReportItemStatus.INVALID)));
+            }
+        }
+
+        [NUnit.Framework.Test]
+        public virtual void TaggedDocModifyAnnotationTest() {
+            using (PdfDocument document = new PdfDocument(new PdfReader(SOURCE_FOLDER + "taggedDocModifyAnnotation.pdf"
+                ))) {
+                DocumentRevisionsValidator validator = builder.BuildDocumentRevisionsValidator();
+                ValidationReport report = validator.ValidateAllDocumentRevisions(validationContext, document);
+                AssertValidationReport.AssertThat(report, (a) => a.HasStatus(ValidationReport.ValidationResult.VALID).HasNumberOfLogs
+                    (0));
+            }
+        }
+
+        [NUnit.Framework.Test]
+        public virtual void TaggedDocModifyAnnotationAndStructElementTest() {
+            using (PdfDocument document = new PdfDocument(new PdfReader(SOURCE_FOLDER + "taggedDocModifyAnnotationAndStructElement.pdf"
+                ))) {
+                DocumentRevisionsValidator validator = builder.BuildDocumentRevisionsValidator();
+                ValidationReport report = validator.ValidateAllDocumentRevisions(validationContext, document);
+                AssertValidationReport.AssertThat(report, (a) => a.HasStatus(ValidationReport.ValidationResult.INVALID).HasNumberOfFailures
+                    (2).HasLogItem((l) => l.WithCheckName(DocumentRevisionsValidator.DOC_MDP_CHECK).WithMessage(DocumentRevisionsValidator
+                    .STRUCT_TREE_ROOT_MODIFIED).WithStatus(ReportItem.ReportItemStatus.INVALID)).HasLogItem((l) => l.WithCheckName
+                    (DocumentRevisionsValidator.DOC_MDP_CHECK).WithMessage(DocumentRevisionsValidator.STRUCT_TREE_ELEMENT_MODIFIED
+                    ).WithStatus(ReportItem.ReportItemStatus.INVALID)));
+            }
+        }
+
+        [NUnit.Framework.Test]
+        public virtual void TaggedDocModifyAnnotationAndStructContentTest() {
+            using (PdfDocument document = new PdfDocument(new PdfReader(SOURCE_FOLDER + "taggedDocModifyAnnotationAndStructContent.pdf"
+                ))) {
+                DocumentRevisionsValidator validator = builder.BuildDocumentRevisionsValidator();
+                ValidationReport report = validator.ValidateAllDocumentRevisions(validationContext, document);
+                AssertValidationReport.AssertThat(report, (a) => a.HasStatus(ValidationReport.ValidationResult.VALID).HasNumberOfLogs
+                    (0));
+            }
+        }
+
+        [NUnit.Framework.Test]
+        public virtual void TaggedDocModifyStructElementTest() {
+            using (PdfDocument document = new PdfDocument(new PdfReader(SOURCE_FOLDER + "taggedDocModifyStructElement.pdf"
+                ))) {
+                DocumentRevisionsValidator validator = builder.BuildDocumentRevisionsValidator();
+                ValidationReport report = validator.ValidateAllDocumentRevisions(validationContext, document);
+                AssertValidationReport.AssertThat(report, (a) => a.HasStatus(ValidationReport.ValidationResult.INVALID).HasNumberOfFailures
+                    (2).HasLogItem((l) => l.WithCheckName(DocumentRevisionsValidator.DOC_MDP_CHECK).WithMessage(DocumentRevisionsValidator
+                    .STRUCT_TREE_ROOT_MODIFIED).WithStatus(ReportItem.ReportItemStatus.INVALID)).HasLogItem((l) => l.WithCheckName
+                    (DocumentRevisionsValidator.DOC_MDP_CHECK).WithMessage(DocumentRevisionsValidator.STRUCT_TREE_ELEMENT_MODIFIED
+                    ).WithStatus(ReportItem.ReportItemStatus.INVALID)));
+            }
+        }
+
         [NUnit.Framework.Test]
         public virtual void RemoveUnnamedFieldTest() {
             using (PdfDocument document = new PdfDocument(new PdfReader(SOURCE_FOLDER + "removeUnnamedField.pdf"))) {

itext.tests/itext.sign.tests/itext/signatures/validation/v1/DocumentRevisionsValidatorTest.cs

@@ -134,8 +134,8 @@ public virtual void ExtensionsModificationsTest() {
                     );
                 AssertValidationReport.AssertThat(validationReport, (a) => a.HasStatus(ValidationReport.ValidationResult.INVALID
                     ).HasNumberOfFailures(1).HasNumberOfLogs(1).HasLogItem((l) => l.WithCheckName(DocumentRevisionsValidator
-                    .DOC_MDP_CHECK).WithMessage(DocumentRevisionsValidator.DEVELOPER_EXTENSION_REMOVED, (i) => PdfName.ESIC
-                    ).WithStatus(ReportItem.ReportItemStatus.INVALID)));
+                    .DOC_MDP_CHECK).WithMessage(DocumentRevisionsValidator.BASE_VERSION_DECREASED, (i) => PdfName.ESIC).WithStatus
+                    (ReportItem.ReportItemStatus.INVALID)));
                 validationReport = new ValidationReport();
                 validator.ValidateRevision(documentRevisions[2], documentRevisions[3], document, validationReport, validationContext
                     );