Streaming Rückbau

git-svn-id: https://joinup.ec.europa.eu/svn/pdf-as/trunk@210 7b5415b0-85f9-ee4d-85bd-d5d0c3b42d1c

10 files changed, 175 insertions, 21 deletions

diff --git a/src/main/java/at/gv/egiz/pdfas/impl/input/ByteArrayPdfDataSourceImpl.java b/src/main/java/at/gv/egiz/pdfas/impl/input/ByteArrayPdfDataSourceImpl.java
index 0d27781..e4a5649 100644
--- a/src/main/java/at/gv/egiz/pdfas/impl/input/ByteArrayPdfDataSourceImpl.java
+++ b/src/main/java/at/gv/egiz/pdfas/impl/input/ByteArrayPdfDataSourceImpl.java
@@ -23,34 +23,43 @@ public class ByteArrayPdfDataSourceImpl implements PdfDataSource
 {

   protected byte[] pdf = null;

   

-  protected int length = -1;

-

   public ByteArrayPdfDataSourceImpl(byte[] pdf)

   {

     PerformanceCounters.byteArrays.increment();

 

     this.pdf = pdf;

-    this.length = pdf.length;

   }

 

   public ByteArrayPdfDataSourceImpl(byte[] pdf, int length)

   {

     PerformanceCounters.byteArrays.increment();

 

-    this.pdf = pdf;

-    this.length = length;

+    if (pdf.length == length)

+    {

+      this.pdf = pdf;

+    }

+    else

+    {

+      this.pdf = new byte [length];

+      System.arraycopy(pdf, 0, this.pdf, 0, length);

+    }

   }

 

 

   public InputStream createInputStream()

   {

-    ByteArrayInputStream bais = new ByteArrayInputStream(this.pdf, 0, this.length);

+    ByteArrayInputStream bais = new ByteArrayInputStream(this.pdf);

     return bais;

   }

 

   public int getLength()

   {

-    return this.length;

+    return this.pdf.length;

   }

 

+  public byte[] getAsByteArray()

+  {

+    return this.pdf;

+  }

+  

 }

diff --git a/src/main/java/at/gv/egiz/pdfas/impl/input/CompoundPdfDataSourceImpl.java b/src/main/java/at/gv/egiz/pdfas/impl/input/CompoundPdfDataSourceImpl.java
index f77d6be..4f00bb5 100644
--- a/src/main/java/at/gv/egiz/pdfas/impl/input/CompoundPdfDataSourceImpl.java
+++ b/src/main/java/at/gv/egiz/pdfas/impl/input/CompoundPdfDataSourceImpl.java
@@ -43,5 +43,23 @@ public class CompoundPdfDataSourceImpl implements PdfDataSource
   {

     return this.originalDataSource.getLength() + this.appendix.length;

   }

+  

+  byte [] cache = null;

 

+  /**

+   * @see at.gv.egiz.pdfas.framework.input.DataSource#getAsByteArray()

+   */

+  public byte[] getAsByteArray()

+  {

+    if (cache != null)

+    {

+      return cache;

+    }

+    

+    cache = new byte [getLength()];

+    System.arraycopy(originalDataSource.getAsByteArray(), 0, cache, 0, originalDataSource.getLength());

+    System.arraycopy(appendix, 0, cache, originalDataSource.getLength(), appendix.length);

+    

+    return cache;

+  }

 }

diff --git a/src/main/java/at/gv/egiz/pdfas/impl/input/DelimitedPdfDataSource.java b/src/main/java/at/gv/egiz/pdfas/impl/input/DelimitedPdfDataSource.java
index 6c67be2..63874c7 100644
--- a/src/main/java/at/gv/egiz/pdfas/impl/input/DelimitedPdfDataSource.java
+++ b/src/main/java/at/gv/egiz/pdfas/impl/input/DelimitedPdfDataSource.java
@@ -41,4 +41,22 @@ public class DelimitedPdfDataSource implements PdfDataSource
     return this.len;

   }

 

+  byte [] cache = null;

+

+  /**

+   * @see at.gv.egiz.pdfas.framework.input.DataSource#getAsByteArray()

+   */

+  public byte[] getAsByteArray()

+  {

+    if (cache != null)

+    {

+      return cache;

+    }

+    

+    cache = new byte [getLength()];

+    System.arraycopy(dataSource.getAsByteArray(), 0, cache, 0, getLength());

+    

+    return cache;

+  }

+

 }

diff --git a/src/main/java/at/gv/egiz/pdfas/impl/input/FileBasedPdfDataSourceImpl.java b/src/main/java/at/gv/egiz/pdfas/impl/input/FileBasedPdfDataSourceImpl.java
index 8453192..a3a0803 100644
--- a/src/main/java/at/gv/egiz/pdfas/impl/input/FileBasedPdfDataSourceImpl.java
+++ b/src/main/java/at/gv/egiz/pdfas/impl/input/FileBasedPdfDataSourceImpl.java
@@ -3,6 +3,7 @@
  */

 package at.gv.egiz.pdfas.impl.input;

 

+import java.io.ByteArrayInputStream;

 import java.io.File;

 import java.io.FileInputStream;

 import java.io.FileNotFoundException;

@@ -13,6 +14,7 @@ import org.apache.commons.logging.Log;
 import org.apache.commons.logging.LogFactory;

 

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

+import at.gv.egiz.pdfas.impl.input.helper.DataSourceHelper;

 

 /**

  * @author wprinz

@@ -78,6 +80,15 @@ public class FileBasedPdfDataSourceImpl implements PdfDataSource, FileBased
    */

   public InputStream createInputStream()

   {

+    if (cache == null)

+    {

+      getAsByteArray();

+    }

+    return new ByteArrayInputStream(cache);

+  }

+

+  protected InputStream createFileInputStream()

+  {

     try

     {

       FileInputStream fis = new FileInputStream(getFile());

@@ -92,7 +103,6 @@ public class FileBasedPdfDataSourceImpl implements PdfDataSource, FileBased
       return null;

     }

   }

-

   /**

    * @see at.gv.egiz.pdfas.framework.input.PdfDataSource#getLength()

    */

@@ -100,4 +110,22 @@ public class FileBasedPdfDataSourceImpl implements PdfDataSource, FileBased
   {

     return this.length;

   }

+  

+  byte [] cache = null;

+  

+  /**

+   * @see at.gv.egiz.pdfas.framework.input.DataSource#getAsByteArray()

+   */

+  public byte[] getAsByteArray()

+  {

+    if (cache != null)

+    {

+      return cache;

+    }

+    

+    cache = DataSourceHelper.convertInputStreamToByteArray(createFileInputStream());

+    

+    return cache;

+  }

+

 }

diff --git a/src/main/java/at/gv/egiz/pdfas/impl/input/FileBasedTextDataSourceImpl.java b/src/main/java/at/gv/egiz/pdfas/impl/input/FileBasedTextDataSourceImpl.java
index 6f6c7b4..1988519 100644
--- a/src/main/java/at/gv/egiz/pdfas/impl/input/FileBasedTextDataSourceImpl.java
+++ b/src/main/java/at/gv/egiz/pdfas/impl/input/FileBasedTextDataSourceImpl.java
@@ -13,6 +13,7 @@ import org.apache.commons.logging.Log;
 import org.apache.commons.logging.LogFactory;

 

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

+import at.gv.egiz.pdfas.impl.input.helper.DataSourceHelper;

 

 /**

  * @author wprinz

@@ -121,4 +122,21 @@ public class FileBasedTextDataSourceImpl implements TextDataSource, FileBased
     return (int) getFile().length();

   }

 

+  byte [] cache = null;

+  

+  /**

+   * @see at.gv.egiz.pdfas.framework.input.DataSource#getAsByteArray()

+   */

+  public byte[] getAsByteArray()

+  {

+    if (cache != null)

+    {

+      return cache;

+    }

+    

+    cache = DataSourceHelper.convertInputStreamToByteArray(createInputStream());

+    

+    return cache;

+  }

+

 }

diff --git a/src/main/java/at/gv/egiz/pdfas/impl/input/TextDataSourceImpl.java b/src/main/java/at/gv/egiz/pdfas/impl/input/TextDataSourceImpl.java
index b259a3e..c983a8a 100644
--- a/src/main/java/at/gv/egiz/pdfas/impl/input/TextDataSourceImpl.java
+++ b/src/main/java/at/gv/egiz/pdfas/impl/input/TextDataSourceImpl.java
@@ -79,4 +79,20 @@ public class TextDataSourceImpl implements TextDataSource
     }

   }

 

+  /**

+   * @see at.gv.egiz.pdfas.framework.input.DataSource#getAsByteArray()

+   */

+  public byte[] getAsByteArray()

+  {

+    try

+    {

+      byte[] data = getText().getBytes("UTF-8");

+      return data;

+    }

+    catch (UnsupportedEncodingException e)

+    {

+      throw new RuntimeException(e);

+    }

+  }

+

 }

diff --git a/src/main/java/at/gv/egiz/pdfas/impl/input/helper/DataSourceHelper.java b/src/main/java/at/gv/egiz/pdfas/impl/input/helper/DataSourceHelper.java
index 1e2ffdc..138b269 100644
--- a/src/main/java/at/gv/egiz/pdfas/impl/input/helper/DataSourceHelper.java
+++ b/src/main/java/at/gv/egiz/pdfas/impl/input/helper/DataSourceHelper.java
@@ -3,6 +3,7 @@
  */

 package at.gv.egiz.pdfas.impl.input.helper;

 

+import java.io.ByteArrayOutputStream;

 import java.io.File;

 import java.io.FileOutputStream;

 import java.io.IOException;

@@ -42,24 +43,54 @@ public class DataSourceHelper
    */

   public static byte[] convertDataSourceToByteArray(DataSource pdfDataSource)

   {

+    return pdfDataSource.getAsByteArray();

+//    try

+//    {

+//      PerformanceCounters.byteArrays.increment();

+//

+//      byte[] data = new byte[pdfDataSource.getLength()];

+//

+//      int bytes_written = 0;

+//

+//      InputStream is = pdfDataSource.createInputStream();

+//      int n = 0;

+//      while ((n = is.read(data, bytes_written, data.length - bytes_written)) > 0)

+//      {

+//        bytes_written += n;

+//      }

+//      is.close();

+//

+//      assert bytes_written == data.length;

+//

+//      return data;

+//    }

+//    catch (IOException e)

+//    {

+//      log.error(e);

+//      throw new RuntimeException(e);

+//    }

+  }

+  

+  public static byte [] convertInputStreamToByteArray(InputStream inputStream)

+  {

     try

     {

       PerformanceCounters.byteArrays.increment();

 

-      byte[] data = new byte[pdfDataSource.getLength()];

+      ByteArrayOutputStream baos = new ByteArrayOutputStream(4096);

+      

+      byte[] temp = new byte[4096];

 

-      int bytes_written = 0;

-

-      InputStream is = pdfDataSource.createInputStream();

       int n = 0;

-      while ((n = is.read(data, bytes_written, data.length - bytes_written)) > 0)

+      while ((n = inputStream.read(temp)) > 0)

       {

-        bytes_written += n;

+        baos.write(temp, 0, n);

       }

-      is.close();

-

-      assert bytes_written == data.length;

+      inputStream.close();

 

+      baos.close();

+      byte [] data = baos.toByteArray();

+      

       return data;

     }

     catch (IOException e)

diff --git a/src/main/java/at/gv/egiz/pdfas/impl/output/ByteArrayDataSink.java b/src/main/java/at/gv/egiz/pdfas/impl/output/ByteArrayDataSink.java
index f3d1283..3a3aeca 100644
--- a/src/main/java/at/gv/egiz/pdfas/impl/output/ByteArrayDataSink.java
+++ b/src/main/java/at/gv/egiz/pdfas/impl/output/ByteArrayDataSink.java
@@ -47,6 +47,9 @@ public class ByteArrayDataSink implements DataSink
    */

   public OutputStream createOutputStream(String mimeType, String characterEncoding)

   {

+    this.mimeType = mimeType;

+    this.characterEncoding = characterEncoding;

+    

     if (this.baos != null)

     {

       log.warn("An output stream is created twice. The old one will be rendered useless.");

diff --git a/src/main/java/at/gv/egiz/pdfas/impl/signator/textual/TextualSignator_1_0_0.java b/src/main/java/at/gv/egiz/pdfas/impl/signator/textual/TextualSignator_1_0_0.java
index b0494c8..1714afc 100644
--- a/src/main/java/at/gv/egiz/pdfas/impl/signator/textual/TextualSignator_1_0_0.java
+++ b/src/main/java/at/gv/egiz/pdfas/impl/signator/textual/TextualSignator_1_0_0.java
@@ -91,7 +91,7 @@ public class TextualSignator_1_0_0 implements Signator
       tsi.profile = profile;

       tsi.pos = pos;

 

-      String document_text = PdfAS.extractNormalizedTextTextual(pdfDataSource.createInputStream());

+      String document_text = PdfAS.extractNormalizedTextTextual(pdfDataSource);

       // logger_.debug("signed_text = " + document_text);

 

       DataSource ds = new TextDataSourceImpl(document_text);

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 981b868..0c9e1f2 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
@@ -18,7 +18,7 @@ import at.gv.egiz.pdfas.framework.input.TextDataSource;
 import at.gv.egiz.pdfas.framework.verificator.Verificator;

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

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

-import at.gv.egiz.pdfas.impl.input.DelimitedInputStream;

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

 import at.gv.egiz.pdfas.impl.input.helper.DataSourceHelper;

 import at.gv.egiz.pdfas.impl.vfilter.helper.VerificationFilterBinaryHelper;

 import at.gv.egiz.pdfas.impl.vfilter.helper.VerificationFilterHelper;

@@ -285,8 +285,9 @@ public class VerificationFilterImpl implements VerificationFilter
 

   protected String extractText(PdfDataSource pdf, int endOfDocument) throws PresentableException

   {

-    DelimitedInputStream dis = new DelimitedInputStream(pdf.createInputStream(), endOfDocument);

-    return PdfAS.extractNormalizedTextTextual(dis);

+    DelimitedPdfDataSource dds = new DelimitedPdfDataSource(pdf, endOfDocument);

+    //DelimitedInputStream dis = new DelimitedInputStream(pdf.createInputStream(), endOfDocument);

+    return PdfAS.extractNormalizedTextTextual(dds);

   }

 

   protected List extractNewSignaturesFromText(String text) throws VerificationFilterException

@@ -453,6 +454,18 @@ public class VerificationFilterImpl implements VerificationFilter
       log.debug("Extracting signatures:");

       List extractedSignatures = extractNewSignaturesFromText(extractedText);

       log.debug("Extracting signatures finished.");

+      

+      if (log.isDebugEnabled())

+      {

+        log.debug("extracted signatures:");

+        for (int i = 0; i < extractedSignatures.size(); i++)

+        {

+          SignatureHolder sh = (SignatureHolder)extractedSignatures.get(i);

+          String dateStr = sh.getSignatureObject().getSignationDate();

+          EGIZDate ed = EGIZDate.parseFromString(dateStr);

+          log.debug("#" + i + ": dateStr = " + dateStr + ", egizDate = " + ed.toString());

+        }

+      }

 

       return extractedSignatures;

     }