improved nonObjectExtraction. Use per signature basis

git-svn-id: https://joinup.ec.europa.eu/svn/pdf-as/trunk@506 7b5415b0-85f9-ee4d-85bd-d5d0c3b42d1c

11 files changed, 126 insertions, 110 deletions

diff --git a/src/main/java/at/gv/egiz/pdfas/api/analyze/AnalyzeResult.java b/src/main/java/at/gv/egiz/pdfas/api/analyze/AnalyzeResult.java
index c9dde9e..a1aabb4 100644
--- a/src/main/java/at/gv/egiz/pdfas/api/analyze/AnalyzeResult.java
+++ b/src/main/java/at/gv/egiz/pdfas/api/analyze/AnalyzeResult.java
@@ -30,17 +30,5 @@ public interface AnalyzeResult
   

   public List getNoSignatures();

 

-  /**

-   * Returns non textual data from pdf provided that {@link AnalyzeParameters#setReturnNonTextualObjects(boolean)}

-   * has been set {@code true}. 

-   * @return List of {@link NonTextObjectInfo}

-   */

-  public List getNonTextualObjects();

-

-  /**

-   * Returns <code>true</code> if non textual objects have been found, <code>false</code> if not.

-   * @return <code>true</code> if non textual objects have been found, <code>false</code> if not.

-   */

-  public boolean hasNonTextualObjects();

   

 }

diff --git a/src/main/java/at/gv/egiz/pdfas/api/commons/SignatureInformation.java b/src/main/java/at/gv/egiz/pdfas/api/commons/SignatureInformation.java
index 8eef6e8..3b05685 100644
--- a/src/main/java/at/gv/egiz/pdfas/api/commons/SignatureInformation.java
+++ b/src/main/java/at/gv/egiz/pdfas/api/commons/SignatureInformation.java
@@ -19,7 +19,9 @@ package at.gv.egiz.pdfas.api.commons;
 

 import java.security.cert.X509Certificate;

 import java.util.Date;

+import java.util.List;

 

+import at.gv.egiz.pdfas.api.analyze.NonTextObjectInfo;

 import at.gv.egiz.pdfas.api.io.DataSource;

 

 /**

@@ -99,5 +101,21 @@ public interface SignatureInformation
    * @return 

    */

   public String getTimeStampValue();

+  

+  /**

+   * Returns a list<{@link NonTextObjectInfo}> of non textual objects in the pdf document.

+   * Only available for textual signatures. Show this to the user who signed the textual content only! 

+   * @return  List<{@link NonTextObjectInfo} or <tt>null</tt> of not available (binary signature)

+   */

+  public List getNonTextualObjects();

+  

+  public void setNonTextualObjects(List nonTextualObjects);

+

+  /**

+   * Returns <code>true</code> if non textual objects have been found, <code>false</code> if not.

+   * @return <code>true</code> if non textual objects have been found, <code>false</code> if not.

+   */

+  public boolean hasNonTextualObjects();

+

 

 }
\ No newline at end of file
diff --git a/src/main/java/at/gv/egiz/pdfas/api/verify/VerifyResult.java b/src/main/java/at/gv/egiz/pdfas/api/verify/VerifyResult.java
index 34fbcd1..9455a2d 100644
--- a/src/main/java/at/gv/egiz/pdfas/api/verify/VerifyResult.java
+++ b/src/main/java/at/gv/egiz/pdfas/api/verify/VerifyResult.java
@@ -3,6 +3,7 @@ package at.gv.egiz.pdfas.api.verify;
 import java.util.Date;

 import java.util.List;

 

+import at.gv.egiz.pdfas.api.analyze.NonTextObjectInfo;

 import at.gv.egiz.pdfas.api.commons.SignatureInformation;

 

 /**

@@ -113,6 +114,20 @@ public interface VerifyResult extends SignatureInformation
    * @see SignatureInformation#getSignedData()

    */

   public String getHashInputData();

+  

+  /**

+   * Returns a list<{@link NonTextObjectInfo}> of non textual objects in the pdf document.

+   * Only available for textual signatures. Show this to the user who signed the textual content only! 

+   * @return  List<{@link NonTextObjectInfo} or <tt>null</tt> of not available (binary signature)

+   */

+  public List getNonTextualObjects();

+  

+

+  /**

+   * Returns <code>true</code> if non textual objects have been found, <code>false</code> if not.

+   * @return <code>true</code> if non textual objects have been found, <code>false</code> if not.

+   */

+  public boolean hasNonTextualObjects();

 

 

 }

diff --git a/src/main/java/at/gv/egiz/pdfas/api/verify/VerifyResults.java b/src/main/java/at/gv/egiz/pdfas/api/verify/VerifyResults.java
index ef06880..d2f79b3 100644
--- a/src/main/java/at/gv/egiz/pdfas/api/verify/VerifyResults.java
+++ b/src/main/java/at/gv/egiz/pdfas/api/verify/VerifyResults.java
@@ -5,8 +5,6 @@ package at.gv.egiz.pdfas.api.verify;
 

 import java.util.List;

 

-import at.gv.egiz.pdfas.api.analyze.NonTextObjectInfo;

-

 /**

  * The result of the verification of a document.

  * 

@@ -26,19 +24,5 @@ public interface VerifyResults
    *         signature.

    */

   public List getResults();

-  

-  /**

-   * Returns a list<{@link NonTextObjectInfo}> of non textual objects in the pdf document.

-   * Only available for textual signatures. Show this to the user who signed the textual content only! 

-   * @return  List<{@link NonTextObjectInfo} or <tt>null</tt> of not available (binary signature)

-   */

-  public List getNonTextualObjects();

-  

-

-  /**

-   * Returns <code>true</code> if non textual objects have been found, <code>false</code> if not.

-   * @return <code>true</code> if non textual objects have been found, <code>false</code> if not.

-   */

-  public boolean hasNonTextualObjects();

-  

+    

 }

diff --git a/src/main/java/at/gv/egiz/pdfas/impl/api/PdfAsObject.java b/src/main/java/at/gv/egiz/pdfas/impl/api/PdfAsObject.java
index f98b0f0..824416f 100644
--- a/src/main/java/at/gv/egiz/pdfas/impl/api/PdfAsObject.java
+++ b/src/main/java/at/gv/egiz/pdfas/impl/api/PdfAsObject.java
@@ -31,6 +31,7 @@ import at.gv.egiz.pdfas.commandline.CommandlineConnectorChooser;
 import at.gv.egiz.pdfas.exceptions.ErrorCode;

 import at.gv.egiz.pdfas.framework.config.SettingsHelper;

 import at.gv.egiz.pdfas.framework.input.ExtractionStage;

+import at.gv.egiz.pdfas.framework.input.PdfDataSource;

 import at.gv.egiz.pdfas.framework.signator.SignatorInformation;

 import at.gv.egiz.pdfas.framework.vfilter.VerificationFilterParameters;

 import at.gv.egiz.pdfas.impl.api.analyze.AnalyzeResultImpl;

@@ -43,6 +44,7 @@ import at.gv.egiz.pdfas.impl.api.sign.ActualSignaturePositionAdapter;
 import at.gv.egiz.pdfas.impl.api.sign.SignResultImpl;

 import at.gv.egiz.pdfas.impl.api.verify.VerifyResultAdapter;

 import at.gv.egiz.pdfas.impl.api.verify.VerifyResultsImpl;

+import at.gv.egiz.pdfas.impl.input.DelimitedPdfDataSource;

 import at.gv.egiz.pdfas.impl.vfilter.VerificationFilterParametersImpl;

 import at.gv.egiz.pdfas.utils.ConfigUtils;

 import at.knowcenter.wag.egov.egiz.PdfAS;

@@ -52,11 +54,12 @@ import at.knowcenter.wag.egov.egiz.exceptions.PDFDocumentException;
 import at.knowcenter.wag.egov.egiz.exceptions.PresentableException;

 import at.knowcenter.wag.egov.egiz.exceptions.SignatureException;

 import at.knowcenter.wag.egov.egiz.framework.SignatorFactory;

+import at.knowcenter.wag.egov.egiz.pdf.BinarySignatureHolder;

 import at.knowcenter.wag.egov.egiz.pdf.NoSignatureHolder;

 import at.knowcenter.wag.egov.egiz.pdf.ObjectExtractor;

 import at.knowcenter.wag.egov.egiz.pdf.SignatureHolder;

 import at.knowcenter.wag.egov.egiz.pdf.TablePos;

-import at.knowcenter.wag.egov.egiz.pdf.TextualSignature;

+import at.knowcenter.wag.egov.egiz.pdf.TextualSignatureHolder;

 import at.knowcenter.wag.egov.egiz.sig.SignatureResponse;

 import at.knowcenter.wag.egov.egiz.sig.SignatureTypeDefinition;

 import at.knowcenter.wag.egov.egiz.sig.SignatureTypes;

@@ -378,17 +381,35 @@ public class PdfAsObject implements PdfAs
       

         SignatureInformation si = new SignatureInformationAdapter(sh);

         sigInfs.add(si);

+        if (analyzeParameters.isReturnNonTextualObjects()) {           

+           si.setNonTextualObjects(doExtractNonTexualObjects(sh, (PdfDataSource) inputDataSource));           

+        }        

+        

       }

     }

-    List nonTextInfos = null;

-    if (analyzeParameters.isReturnNonTextualObjects()) {

-       nonTextInfos = ObjectExtractor.extractNonTextInfo(new PdfDataSourceAdapter(analyzeParameters.getDocument()));

-    }

     

-    return new AnalyzeResultImpl(sigInfs, noSigs, nonTextInfos);

+    return new AnalyzeResultImpl(sigInfs, noSigs);

   }

 

-  /**

+  private List doExtractNonTexualObjects(SignatureHolder sh, PdfDataSource pdfDataSource) {

+     if (sh == null) return null;

+     if (sh instanceof BinarySignatureHolder) {

+        BinarySignatureHolder bsh = (BinarySignatureHolder)sh;

+        return ObjectExtractor.extractNonTextInfo(bsh.getSignedPdf());

+     } else if (sh instanceof TextualSignatureHolder) {

+        TextualSignatureHolder tsh = (TextualSignatureHolder)sh;

+        if (tsh.getUiBlockEndPos() == 0) {

+           log.warn("uiblockendpos not available. Extract objects from final pdf document");

+           return ObjectExtractor.extractNonTextInfo(pdfDataSource);

+        }

+        DelimitedPdfDataSource dpds = new DelimitedPdfDataSource(pdfDataSource, tsh.getUiBlockEndPos());

+        return ObjectExtractor.extractNonTextInfo(dpds);

+     } else {

+        return null;

+     }

+  }

+

+/**

    * @see at.gv.egiz.pdfas.api.PdfAs#verify(at.gv.egiz.pdfas.api.verify.VerifyAfterAnalysisParameters)

    */

   public VerifyResults verify(VerifyAfterAnalysisParameters verifyAfterAnalysisParameters) throws PdfAsException

@@ -424,11 +445,12 @@ public class PdfAsObject implements PdfAs
       SignatureHolder holder = (SignatureHolder) signature_holders.get(i);

 

       VerifyResult vr = new VerifyResultAdapter(response, holder, verifyAfterAnalysisParameters.getVerificationTime());

+      vr.setNonTextualObjects( ((SignatureInformation) signatures.get(i)).getNonTextualObjects());

+

       vrs.add(vr);

     }

     

     VerifyResultsImpl verifyResults = new VerifyResultsImpl(vrs); 

-    verifyResults.setNonTextualObjects(verifyAfterAnalysisParameters.getAnalyzeResult().getNonTextualObjects());

     return verifyResults;

   }

 

diff --git a/src/main/java/at/gv/egiz/pdfas/impl/api/analyze/AnalyzeResultImpl.java b/src/main/java/at/gv/egiz/pdfas/impl/api/analyze/AnalyzeResultImpl.java
index a50b7a1..154d73d 100644
--- a/src/main/java/at/gv/egiz/pdfas/impl/api/analyze/AnalyzeResultImpl.java
+++ b/src/main/java/at/gv/egiz/pdfas/impl/api/analyze/AnalyzeResultImpl.java
@@ -5,9 +5,7 @@ package at.gv.egiz.pdfas.impl.api.analyze;
 

 import java.util.List;

 

-import at.gv.egiz.pdfas.api.analyze.AnalyzeParameters;

 import at.gv.egiz.pdfas.api.analyze.AnalyzeResult;

-import at.gv.egiz.pdfas.api.analyze.NonTextObjectInfo;

 import at.gv.egiz.pdfas.api.exceptions.PdfAsException;

 

 /**

@@ -25,23 +23,8 @@ public class AnalyzeResultImpl implements AnalyzeResult
   /**

    * The found non-signature update blocks.

    */

-  protected List noSignatures = null;

+  protected List noSignatures = null;  

   

-  protected List nonTextualObjects = null;

-  

-

-  /**

-   * Constructor.

-   * 

-   * @param signatures

-   *          The found signatures.

-   * @param noSignatures

-   *          The found non-signature update blocks.             

-   */

-  public AnalyzeResultImpl(List signatures, List noSignatures)

-  {

-     this(signatures, noSignatures, null);

-  }

 

   /**

    * Constructor.

@@ -50,9 +33,8 @@ public class AnalyzeResultImpl implements AnalyzeResult
    *          The found signatures.

    * @param noSignatures

    *          The found non-signature update blocks.

-   * @param nonTextInfos Non textual data from pdf provided that {@link AnalyzeParameters#setReturnNonTextualObjects(boolean)} was set.              

    */

-  public AnalyzeResultImpl(List signatures, List noSignatures, List nonTextInfos)

+  public AnalyzeResultImpl(List signatures, List noSignatures)

   {

     if (signatures == null)

     {

@@ -60,8 +42,7 @@ public class AnalyzeResultImpl implements AnalyzeResult
     }

 

     this.signatures = signatures;

-    this.noSignatures = noSignatures;

-    this.nonTextualObjects = nonTextInfos;

+    this.noSignatures = noSignatures;   

   }

   

 

@@ -94,22 +75,4 @@ public class AnalyzeResultImpl implements AnalyzeResult
     return this.noSignatures;

   }

 

-

-  /**

-   * Returns non textual data from pdf provided that {@link AnalyzeParameters#setReturnNonTextualObjects(boolean)}

-   * has been set {@code true}. 

-   * @return List of {@link NonTextObjectInfo}

-   */

-  public List getNonTextualObjects() {

-     return this.nonTextualObjects;

-  }

-

-  /**

-   * Returns <code>true</code> if non textual objects have been found, <code>false</code> if not.

-   * @return <code>true</code> if non textual objects have been found, <code>false</code> if not.

-   */

-  public boolean hasNonTextualObjects() {

-   return this.nonTextualObjects != null && !this.nonTextualObjects.isEmpty();

-  }

-  

 }

diff --git a/src/main/java/at/gv/egiz/pdfas/impl/api/commons/SignatureInformationAdapter.java b/src/main/java/at/gv/egiz/pdfas/impl/api/commons/SignatureInformationAdapter.java
index 3d4b560..4661d4e 100644
--- a/src/main/java/at/gv/egiz/pdfas/impl/api/commons/SignatureInformationAdapter.java
+++ b/src/main/java/at/gv/egiz/pdfas/impl/api/commons/SignatureInformationAdapter.java
@@ -5,6 +5,7 @@ package at.gv.egiz.pdfas.impl.api.commons;
 

 import java.security.cert.X509Certificate;

 import java.util.Date;

+import java.util.List;

 

 import at.gv.egiz.pdfas.api.commons.Constants;

 import at.gv.egiz.pdfas.api.commons.SignatureInformation;

@@ -27,6 +28,8 @@ public class SignatureInformationAdapter implements SignatureInformation
   protected SignatureHolder signatureHolder = null;

   

   protected String timeStamp = null;

+  

+  protected List nonTextualObjects = null;

 

   /**

    * Constructor.

@@ -94,5 +97,17 @@ public class SignatureInformationAdapter implements SignatureInformation
       return this.timeStamp;

    }

 

+   public List getNonTextualObjects() {

+      return this.nonTextualObjects;

+   }

+

+   public boolean hasNonTextualObjects() {

+      return this.nonTextualObjects != null && this.nonTextualObjects.size() > 0;

+   }

+

+   public void setNonTextualObjects(List nonTextualObjects) {

+      this.nonTextualObjects = nonTextualObjects;      

+   }

+

   

 }

diff --git a/src/main/java/at/gv/egiz/pdfas/impl/api/verify/VerifyResultsImpl.java b/src/main/java/at/gv/egiz/pdfas/impl/api/verify/VerifyResultsImpl.java
index 22d446a..22ab475 100644
--- a/src/main/java/at/gv/egiz/pdfas/impl/api/verify/VerifyResultsImpl.java
+++ b/src/main/java/at/gv/egiz/pdfas/impl/api/verify/VerifyResultsImpl.java
@@ -15,9 +15,7 @@ public class VerifyResultsImpl implements VerifyResults
   /**

    * The results.

    */

-  protected List results = null;

-  

-  protected List nonTextualObjects = null;

+  protected List results = null;  

 

   /**

    * Constructor.

@@ -38,21 +36,5 @@ public class VerifyResultsImpl implements VerifyResults
     return this.results;

   }

 

-   public List getNonTextualObjects() {

-      return this.nonTextualObjects;

-   }

-

-   public void setNonTextualObjects(List nonTextObjects) {

-      this.nonTextualObjects = nonTextObjects;

-      

-   }

-

-   /**

-    * Returns <code>true</code> if non textual objects have been found, <code>false</code> if not.

-    * @return <code>true</code> if non textual objects have been found, <code>false</code> if not.

-    */

-   public boolean hasNonTextualObjects() {

-      return this.nonTextualObjects != null && !this.nonTextualObjects.isEmpty();

-   }

 

 }

diff --git a/src/main/java/at/gv/egiz/pdfas/impl/vfilter/VerificationFilterImpl.java b/src/main/java/at/gv/egiz/pdfas/impl/vfilter/VerificationFilterImpl.java
index e7724bc..acf6622 100644
--- a/src/main/java/at/gv/egiz/pdfas/impl/vfilter/VerificationFilterImpl.java
+++ b/src/main/java/at/gv/egiz/pdfas/impl/vfilter/VerificationFilterImpl.java
@@ -242,7 +242,7 @@ public class VerificationFilterImpl implements VerificationFilter
       if (sao.oldSignature != null)

       {

         extractedSignatures.add(0, sao.oldSignature);

-      }

+      }      

     }

     else

     {

@@ -666,7 +666,7 @@ public class VerificationFilterImpl implements VerificationFilter
           //kz = new PdfASID("urn:pdfsigfilter:bka.gv.at:binaer:v1.1.0");

           

           Verificator verificator = VerificatorFactory.createBinaryVerificator(kz);

-          List binary_holders = verificator.parseBlock(pdf, data, fpr, prev_end);

+          List binary_holders = verificator.parseBlock(pdf, data, fpr, prev_end);      

 

           binarySignatures.addAll(binary_holders);

           if(binary_holders.size() > 0) {

@@ -753,7 +753,9 @@ public class VerificationFilterImpl implements VerificationFilter
     assert partition.isTextPartition();

 

     int endOfDocument = VerificationFilterHelper.getEndOfPartition(partition);

-    return extractSignatures(pdf, endOfDocument);

+    List extractedSigs = extractSignatures(pdf, endOfDocument);

+    TextualSignatureHolder.mulitSetUiBlockEndPos(extractedSigs, endOfDocument);

+    return extractedSigs;

   }

 

   protected SignaturesAndOld extractSignaturesFromPartitionAndOld(PdfDataSource pdf, Partition partition) throws VerificationFilterException

@@ -770,6 +772,8 @@ public class VerificationFilterImpl implements VerificationFilter
 //      log.debug("extracted text: " + extractedText);

 

       SignaturesAndOld sao = extractSignaturesAndOld(extractedText);

+      TextualSignatureHolder.trySetUiBlockEndPos(sao.oldSignature, endOfDocument);

+      TextualSignatureHolder.mulitSetUiBlockEndPos(sao.newSignatures, endOfDocument);

 

       return sao;

     }

diff --git a/src/main/java/at/knowcenter/wag/egov/egiz/pdf/TextualSignatureHolder.java b/src/main/java/at/knowcenter/wag/egov/egiz/pdf/TextualSignatureHolder.java
index 0971f90..8c7863d 100644
--- a/src/main/java/at/knowcenter/wag/egov/egiz/pdf/TextualSignatureHolder.java
+++ b/src/main/java/at/knowcenter/wag/egov/egiz/pdf/TextualSignatureHolder.java
@@ -18,6 +18,8 @@
 package at.knowcenter.wag.egov.egiz.pdf;

 

 import java.io.Serializable;

+import java.util.Iterator;

+import java.util.List;

 

 import at.gv.egiz.pdfas.framework.input.DataSource;

 import at.gv.egiz.pdfas.framework.input.TextDataSource;

@@ -53,6 +55,8 @@ public class TextualSignatureHolder implements Serializable, SignatureHolder
   private SignatureObject signature_object = null;

   

   private TextDataSource textDataSource = null;

+  

+  private int uiBlockEndPos = 0;

 

   public TextualSignatureHolder(String text, SignatureObject so)

   {

@@ -114,6 +118,28 @@ public class TextualSignatureHolder implements Serializable, SignatureHolder
   {

     return this.textDataSource.getText();

   }

+

+   public int getUiBlockEndPos() {

+      return this.uiBlockEndPos;

+   }

+

+   public void setUiBlockEndPos(int uiBlockEndPos) {

+      this.uiBlockEndPos = uiBlockEndPos;

+   }

+   

+   public static void mulitSetUiBlockEndPos(List signatureHolders, int uiBlockEndPos) {

+      for (Iterator it = signatureHolders.iterator(); it.hasNext();) {

+         SignatureHolder sh = (SignatureHolder) it.next();

+         trySetUiBlockEndPos(sh, uiBlockEndPos);

+      }      

+   }

+   

+   public static void trySetUiBlockEndPos(SignatureHolder sh, int uiBlockEndPos ) {

+      if (sh != null && sh instanceof TextualSignatureHolder) {

+         ((TextualSignatureHolder)sh).setUiBlockEndPos(uiBlockEndPos);

+      }

+   }

+   

   

   

 }

diff --git a/src/main/java/demo/SignatureVerificationDemo.java b/src/main/java/demo/SignatureVerificationDemo.java
index 498849d..6df4973 100644
--- a/src/main/java/demo/SignatureVerificationDemo.java
+++ b/src/main/java/demo/SignatureVerificationDemo.java
@@ -90,19 +90,18 @@ public class SignatureVerificationDemo {
          while (it.hasNext()) {

             VerifyResult result = (VerifyResult) it.next();

             Main.formatVerifyResult(result, out);

-         }

-         out.flush();

-         

-         // check if non textual elements have been detected

-         if (analyzeResult.hasNonTextualObjects()) {

-            System.out.println("\nWARNING: Non textual objects detected.");

-            Iterator noit = analyzeResult.getNonTextualObjects().iterator();

-            while (noit.hasNext()) {

-               NonTextObjectInfo info = (NonTextObjectInfo) noit.next();

-               System.out.println("  -> " + info.toString());

+            if (result.hasNonTextualObjects()) {

+               System.out.println("\nWARNING: Non textual objects detected.");

+               Iterator noit = result.getNonTextualObjects().iterator();

+               while (noit.hasNext()) {

+                  NonTextObjectInfo info = (NonTextObjectInfo) noit.next();

+                  System.out.println("  -> " + info.toString());

+               }

+               System.out.println();

+               

             }

-            System.out.println();

          }

+         out.flush();         

          

       } catch (PdfAsException e) {

          e.printStackTrace();