Fix incorrect check of xref validity which lead to the unnecessary xref rebuilt. Add log message on xref rebuilt

DEVSIX-605

Author: yulian-gaponenko

io/src/main/java/com/itextpdf/io/LogMessageConstant.java

@@ -69,6 +69,8 @@ public final class LogMessageConstant {
      */
     public static final String ELEMENT_DOES_NOT_FIT_AREA = "Element does not fit current area. {0}";
 
+    public static final String EXCEPTION_WHILE_UPDATING_XMPMETADATA = "Exception while updating XmpMetadata";
+
     /**
      * Log message.
      */
@@ -179,5 +181,5 @@ public final class LogMessageConstant {
 
     public static final String WRITER_ENCRYPTION_IS_IGNORED_PRESERVE = "Writer encryption will be ignored, because preservation of encryption is enabled. Document will preserve the original encryption (or will stay unencrypted)";
 
-    public static final String EXCEPTION_WHILE_UPDATING_XMPMETADATA = "Exception while updating XmpMetadata";
+    public static final String XREF_ERROR = "Error occurred while reading cross reference table. Cross reference table will be rebuilt.";
 }

kernel/src/main/java/com/itextpdf/kernel/PdfException.java

@@ -152,8 +152,8 @@ public class PdfException extends RuntimeException {
     public static final String InvalidPageStructurePagesPagesMustBePdfDictionary = "invalid.page.structure.pages.must.be.pdfdictionary";
     public static final String InvalidRangeArray = "invalid.range.array";
     public static final String InvalidOffsetForObject1 = "invalid.offset.for.object {0}";
-    public static final String InvalidXrefSection = "invalid.xref.section";
     public static final String InvalidXrefStream = "invalid.xref.stream";
+    public static final String InvalidXrefTable = "invalid.xref.table";
     public static final String IoException = "io.exception";
     public static final String IsNotAnAcceptableValueForTheField = "{0}.is.not.an.acceptable.value.for.the.field.{1}";
     public static final String IsNotWmfImage = "is.not.wmf.image";

kernel/src/main/java/com/itextpdf/kernel/pdf/PdfIndirectReference.java

@@ -151,19 +151,19 @@ public int getObjStreamNumber() {
     /**
      * Gets refersTo object offset in a document.
      *
-     * @return object offset in a document. If refersTo object is in object stream then 0.
+     * @return object offset in a document. If refersTo object is in object stream then -1.
      */
     public long getOffset() {
-        return objectStreamNumber == 0 ? offsetOrIndex : 0;
+        return objectStreamNumber == 0 ? offsetOrIndex : -1;
     }
 
     /**
      * Gets refersTo object index in the object stream.
      *
-     * @return object index in a document. If refersTo object is not in object stream then 0.
+     * @return object index in a document. If refersTo object is not in object stream then -1.
      */
     public int getIndex() {
-        return objectStreamNumber == 0 ? 0 : (int)offsetOrIndex;
+        return objectStreamNumber == 0 ? -1 : (int)offsetOrIndex;
     }
 
     @Override

kernel/src/main/java/com/itextpdf/kernel/pdf/PdfReader.java

@@ -53,7 +53,6 @@ This file is part of the iText (R) project.
 import com.itextpdf.io.source.RandomAccessSourceFactory;
 import com.itextpdf.io.source.WindowRandomAccessSource;
 import com.itextpdf.kernel.PdfException;
-import com.itextpdf.kernel.pdf.filters.DoNothingFilter;
 import com.itextpdf.kernel.pdf.filters.IFilterHandler;
 import com.itextpdf.kernel.pdf.filters.FilterHandlers;
 
@@ -451,6 +450,9 @@ protected void readPdf() throws IOException {
         try {
             readXref();
         } catch (RuntimeException ex) {
+            Logger logger = LoggerFactory.getLogger(PdfReader.class);
+            logger.error(LogMessageConstant.XREF_ERROR, ex);
+
             rebuildXref();
         }
         readDecryptObj();
@@ -711,14 +713,18 @@ protected void readXref() throws IOException {
             tokens.seek(startxref);
             trailer2 = readXrefSection();
         }
+
+        Integer xrefSize = trailer.getAsInt(PdfName.Size);
+        if (xrefSize == null || xrefSize != pdfDocument.getXref().size()) {
+            throw new PdfException(PdfException.InvalidXrefTable);
+        }
     }
 
     protected PdfDictionary readXrefSection() throws IOException {
         tokens.nextValidToken();
         if (!tokens.tokenValueEqualsTo(PdfTokenizer.Xref))
             tokens.throwError(PdfException.XrefSubsectionNotFound);
         PdfXrefTable xref = pdfDocument.getXref();
-        int end = 0;
         while (true) {
             tokens.nextValidToken();
             if (tokens.tokenValueEqualsTo(PdfTokenizer.Trailer)) {
@@ -732,7 +738,7 @@ protected PdfDictionary readXrefSection() throws IOException {
             if (tokens.getTokenType() != PdfTokenizer.TokenType.Number) {
                 tokens.throwError(PdfException.NumberOfEntriesInThisXrefSubsectionNotFound);
             }
-            end = tokens.getIntValue() + start;
+            int end = tokens.getIntValue() + start;
             for (int num = start; num < end; num++) {
                 tokens.nextValidToken();
                 long pos = tokens.getLongValue();
@@ -764,11 +770,6 @@ protected PdfDictionary readXrefSection() throws IOException {
             }
         }
         PdfDictionary trailer = (PdfDictionary) readObject(false);
-        PdfNumber xrefSize = (PdfNumber) trailer.get(PdfName.Size);
-        if (xrefSize == null || (xrefSize.intValue() != end && end > 0)) {
-            throw new PdfException(PdfException.InvalidXrefSection);
-        }
-
         PdfObject xrs = trailer.get(PdfName.XRefStm);
         if (xrs != null && xrs.getType() == PdfObject.NUMBER) {
             int loc = ((PdfNumber) xrs).intValue();

kernel/src/test/java/com/itextpdf/kernel/pdf/PdfReaderTest.java

@@ -666,6 +666,7 @@ public void pagesTest10() throws IOException {
     }
 
     @Test
+    @LogMessages(messages = @LogMessage(messageTemplate = LogMessageConstant.XREF_ERROR, count = 1))
     public void correctSimpleDoc1() throws IOException {
         String filename = sourceFolder + "correctSimpleDoc1.pdf";
 
@@ -702,6 +703,7 @@ public void correctSimpleDoc2() throws IOException {
     }
 
     @Test
+    @LogMessages(messages = @LogMessage(messageTemplate = LogMessageConstant.XREF_ERROR, count = 1))
     public void correctSimpleDoc3() throws IOException {
         String filename = sourceFolder + "correctSimpleDoc3.pdf";
 
@@ -738,6 +740,7 @@ public void correctSimpleDoc4() throws IOException {
     }
 
     @Test
+    @LogMessages(messages = @LogMessage(messageTemplate = LogMessageConstant.XREF_ERROR, count = 1))
     public void fixPdfTest01() throws IOException {
         String filename = sourceFolder + "OnlyTrailer.pdf";
 
@@ -820,6 +823,7 @@ public void fixPdfTest04() throws IOException {
     }
 
     @Test
+    @LogMessages(messages = @LogMessage(messageTemplate = LogMessageConstant.XREF_ERROR, count = 1))
     public void fixPdfTest05() throws IOException {
         String filename = sourceFolder + "CompressionWrongShift.pdf";
 
@@ -873,6 +877,7 @@ public void fixPdfTest07() throws IOException {
     }
 
     @Test
+    @LogMessages(messages = @LogMessage(messageTemplate = LogMessageConstant.XREF_ERROR, count = 1))
     public void fixPdfTest08() throws IOException {
         String filename = sourceFolder + "XRefSectionWithFreeReferences2.pdf";
 
@@ -898,6 +903,7 @@ public void fixPdfTest08() throws IOException {
     }
 
     @Test
+    @LogMessages(messages = @LogMessage(messageTemplate = LogMessageConstant.XREF_ERROR, count = 1))
     public void fixPdfTest09() throws IOException {
         String filename = sourceFolder + "XRefSectionWithFreeReferences3.pdf";
 
@@ -950,6 +956,7 @@ public void fixPdfTest10() throws IOException {
     }
 
     @Test
+    @LogMessages(messages = @LogMessage(messageTemplate = LogMessageConstant.XREF_ERROR, count = 1))
     public void fixPdfTest11() throws IOException {
         String filename = sourceFolder + "XRefSectionWithoutSize.pdf";
 
@@ -971,6 +978,7 @@ public void fixPdfTest11() throws IOException {
     }
 
     @Test
+    @LogMessages(messages = @LogMessage(messageTemplate = LogMessageConstant.XREF_ERROR, count = 1))
     public void fixPdfTest12() throws IOException {
         String filename = sourceFolder + "XRefWithBreaks.pdf";
 
@@ -1061,6 +1069,7 @@ public void fixPdfTest14() throws IOException {
     }
 
     @Test
+    @LogMessages(messages = @LogMessage(messageTemplate = LogMessageConstant.XREF_ERROR, count = 1))
     public void fixPdfTest15() throws IOException {
         String filename = sourceFolder + "XRefWithInvalidGenerations3.pdf";
 
@@ -1105,6 +1114,7 @@ public void fixPdfTest16() throws IOException {
     }
 
     @Test
+    @LogMessages(messages = @LogMessage(messageTemplate = LogMessageConstant.XREF_ERROR, count = 1))
     public void fixPdfTest17() throws IOException {
         String filename = sourceFolder + "XrefWithNullOffsets.pdf";
 
@@ -1226,6 +1236,7 @@ public void appendModeWith10PagesWithCompression() throws IOException {
     }
 
     @Test
+    @LogMessages(messages = @LogMessage(messageTemplate = LogMessageConstant.XREF_ERROR, count = 1))
     public void appendModeWith10PagesFix1() throws IOException {
         String filename = sourceFolder + "10PagesDocumentAppendedFix1.pdf";
 
@@ -1252,6 +1263,7 @@ public void appendModeWith10PagesFix1() throws IOException {
     }
 
     @Test
+    @LogMessages(messages = @LogMessage(messageTemplate = LogMessageConstant.XREF_ERROR, count = 1))
     public void appendModeWith10PagesFix2() throws IOException {
         String filename = sourceFolder + "10PagesDocumentAppendedFix2.pdf";
 
@@ -1496,6 +1508,22 @@ public void zeroUpdateTest() throws IOException {
         pdfDoc.close();
     }
 
+    @Test
+    public void incrementalUpdateWithOnlyZeroObjectUpdate() throws IOException {
+        String filename = sourceFolder + "pdfReferenceUpdated.pdf";
+
+        FileInputStream fis = new FileInputStream(filename);
+        PdfReader reader = new PdfReader(fis);
+        PdfDocument pdfDoc = new PdfDocument(reader);
+
+        Assert.assertFalse(reader.hasFixedXref());
+        Assert.assertFalse(reader.hasRebuiltXref());
+
+        // problem that is tested here originally was found because the StructTreeRoot dictionary wasn't read
+        Assert.assertTrue(pdfDoc.isTagged());
+        pdfDoc.close();
+    }
+
 
     private boolean objectTypeEqualTo(PdfObject object, PdfName type) {
         PdfName objectType = ((PdfDictionary)object).getAsName(PdfName.Type);