Change Request "Document Correction"

git-svn-id: https://joinup.ec.europa.eu/svn/pdf-as/trunk@239 7b5415b0-85f9-ee4d-85bd-d5d0c3b42d1c

9 files changed, 463 insertions, 9 deletions

diff --git a/src/main/java/at/gv/egiz/pdfas/commandline/Main.java b/src/main/java/at/gv/egiz/pdfas/commandline/Main.java
index 147b540..b4546e4 100644
--- a/src/main/java/at/gv/egiz/pdfas/commandline/Main.java
+++ b/src/main/java/at/gv/egiz/pdfas/commandline/Main.java
@@ -573,10 +573,6 @@ public abstract class Main
 

   public static void processSign(DataSource dataSource, String connector, String signature_mode, String signature_type, String pos_string, DataSink dataSink) throws PdfAsException

   {

-    // FIXME implement this

-    // TODO this will change soon due to the document-cleaning change request.

-    //PdfAS.applyStrictMode(pdfDataSource);

-

     TablePos pos = null;

     if (pos_string != null)

     {

diff --git a/src/main/java/at/gv/egiz/pdfas/exceptions/ErrorCode.java b/src/main/java/at/gv/egiz/pdfas/exceptions/ErrorCode.java
index d2345c8..417c48f 100644
--- a/src/main/java/at/gv/egiz/pdfas/exceptions/ErrorCode.java
+++ b/src/main/java/at/gv/egiz/pdfas/exceptions/ErrorCode.java
@@ -50,6 +50,7 @@ public final class ErrorCode
   public static final int PLACEHOLDER_EXCEPTION = 700;

   public static final int CAPTION_NOT_FOUND_EXCEPTION = 701;

 

-  

+  public static final int CORRECTOR_EXCEPTION = 801;

+  public static final int EXTERNAL_CORRECTOR_TIMEOUT_REACHED = 802;  

 

 }

diff --git a/src/main/java/at/gv/egiz/pdfas/exceptions/framework/CorrectorException.java b/src/main/java/at/gv/egiz/pdfas/exceptions/framework/CorrectorException.java
new file mode 100644
index 0000000..d4759f3
--- /dev/null
+++ b/src/main/java/at/gv/egiz/pdfas/exceptions/framework/CorrectorException.java
@@ -0,0 +1,36 @@
+/**

+ * 

+ */

+package at.gv.egiz.pdfas.exceptions.framework;

+

+import at.knowcenter.wag.egov.egiz.exceptions.PresentableException;

+

+/**

+ * Exceptions thrown by Correctors.

+ * 

+ * @author wprinz

+ * 

+ */

+public class CorrectorException extends PresentableException

+{

+  /**

+   * SVUID.

+   */

+  private static final long serialVersionUID = -8646964226476111797L;

+

+  public CorrectorException(int errorCode, String message, Throwable cause)

+  {

+    super(errorCode, message, cause);

+  }

+

+  public CorrectorException(int errorCode, String message)

+  {

+    super(errorCode, message);

+  }

+

+  public CorrectorException(int errorCode, Throwable cause)

+  {

+    super(errorCode, cause);

+  }

+

+}

diff --git a/src/main/java/at/gv/egiz/pdfas/framework/input/correction/Corrector.java b/src/main/java/at/gv/egiz/pdfas/framework/input/correction/Corrector.java
new file mode 100644
index 0000000..3b4d4b4
--- /dev/null
+++ b/src/main/java/at/gv/egiz/pdfas/framework/input/correction/Corrector.java
@@ -0,0 +1,41 @@
+/**

+ * 

+ */

+package at.gv.egiz.pdfas.framework.input.correction;

+

+import at.gv.egiz.pdfas.exceptions.framework.CorrectorException;

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

+

+/**

+ * Interface for PDF corretors.

+ * 

+ * <p>

+ * Often PDF documents generated by various document to PDF converters have an

+ * invalid structure that upsets PDF-AS. The correction mechanism allows these

+ * documents to be corrected before being signed.

+ * </p>

+ * <p>

+ * A PDF corrector takes an incorrect PDF document and transforms it into a

+ * correct one.

+ * </p>

+ * <p>

+ * Note that correcting a document destroys all signatures in that document, so

+ * never correct an already signed document.

+ * </p>

+ * 

+ * @author wprinz

+ */

+public interface Corrector

+{

+  /**

+   * Corrects the given PDF document to a form that PDF-AS can use.

+   * 

+   * @param document

+   *          The (incorrect) PDF document.

+   * @return Returns the corrected PDF document.

+   * @throws CorrectorException

+   *           Exception thrown if the document couldn't be corrected.

+   */

+  public PdfDataSource correctDocument(PdfDataSource document) throws CorrectorException;

+

+}

diff --git a/src/main/java/at/gv/egiz/pdfas/framework/input/correction/CorrectorFactory.java b/src/main/java/at/gv/egiz/pdfas/framework/input/correction/CorrectorFactory.java
new file mode 100644
index 0000000..083713f
--- /dev/null
+++ b/src/main/java/at/gv/egiz/pdfas/framework/input/correction/CorrectorFactory.java
@@ -0,0 +1,43 @@
+/**

+ * 

+ */

+package at.gv.egiz.pdfas.framework.input.correction;

+

+import at.gv.egiz.pdfas.impl.input.correction.ExternalCorrector;

+import at.gv.egiz.pdfas.impl.input.correction.InternalCorrector;

+import at.knowcenter.wag.egov.egiz.cfg.SettingsReader;

+import at.knowcenter.wag.egov.egiz.exceptions.SettingsException;

+

+/**

+ * Factory for creating Correctors.

+ * 

+ * @author wprinz

+ */

+public class CorrectorFactory

+{

+  public static final String INTERNAL_CORRECTOR = "internal";

+

+  public static final String EXTERNAL_CORRECTOR = "external";

+  

+  public static final String CORRECTOR_KEY = "corrector";

+

+  public static Corrector createCorrector(String id) throws SettingsException

+  {

+    if (id.equals(INTERNAL_CORRECTOR))

+    {

+      return new InternalCorrector();

+    }

+    if (id.equals(EXTERNAL_CORRECTOR))

+    {

+      return new ExternalCorrector();

+    }

+    throw new SettingsException("The connector id '" + id + "' is not a valid corrector id.");

+  }

+

+  public static Corrector createCorrector() throws SettingsException

+  {

+    String id = SettingsReader.getInstance().getSetting(CORRECTOR_KEY, INTERNAL_CORRECTOR);

+    return createCorrector(id);

+  }

+

+}

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 a343de8..3b5d0a6 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
@@ -134,6 +134,8 @@ public class PdfAsObject implements PdfAs
   public SignResult sign(SignParameters signParameters) throws PdfAsException

   {

     CheckHelper.checkSignParameters(signParameters);

+    

+    signParameters.setDocument(PdfAS.applyStrictMode(signParameters.getDocument()));

 

     PdfASID signatorId = null;

     if (signParameters.getSignatureType().equals(Constants.SIGNATURE_TYPE_BINARY))

diff --git a/src/main/java/at/gv/egiz/pdfas/impl/input/correction/ExternalCorrector.java b/src/main/java/at/gv/egiz/pdfas/impl/input/correction/ExternalCorrector.java
new file mode 100644
index 0000000..e244746
--- /dev/null
+++ b/src/main/java/at/gv/egiz/pdfas/impl/input/correction/ExternalCorrector.java
@@ -0,0 +1,263 @@
+/**

+ * 

+ */

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

+

+import java.io.BufferedReader;

+import java.io.File;

+import java.io.IOException;

+import java.io.InputStreamReader;

+

+import org.apache.commons.logging.Log;

+import org.apache.commons.logging.LogFactory;

+

+import at.gv.egiz.pdfas.exceptions.ErrorCode;

+import at.gv.egiz.pdfas.exceptions.framework.CorrectorException;

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

+import at.gv.egiz.pdfas.framework.input.correction.Corrector;

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

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

+import at.gv.egiz.pdfas.web.helper.TempDirHelper;

+import at.knowcenter.wag.egov.egiz.cfg.SettingsReader;

+import at.knowcenter.wag.egov.egiz.exceptions.SettingNotFoundException;

+import at.knowcenter.wag.egov.egiz.exceptions.SettingsException;

+

+/**

+ * Corrects the document using an extrenal commandline tool.

+ * 

+ * <p>

+ * Process.destroy after a certain timeout does not work if the executable is a

+ * Windows batch file.

+ * </p>

+ * 

+ * @author wprinz

+ */

+public class ExternalCorrector implements Corrector

+{

+  public static final String INPUT_DOCUMENT_REPLACE = "##input_document##";

+

+  public static final String OUTPUT_DOCUMENT_REPLACE = "##output_document##";

+

+  public static final String COMMANDLINE_KEY = "external_corrector_commandline";

+

+  public static final String TIMEOUT_KEY = "external_corrector_timeout";

+

+  protected static final int DEFAULT_TIMEOUT = 1000;

+

+  /**

+   * The log.

+   */

+  private static final Log log = LogFactory.getLog(ExternalCorrector.class);

+

+  /**

+   * @see at.gv.egiz.pdfas.framework.input.correction.Corrector#correctDocument(at.gv.egiz.pdfas.framework.input.PdfDataSource)

+   */

+  public PdfDataSource correctDocument(PdfDataSource document) throws CorrectorException

+  {

+

+    try

+    {

+      String outName = null;

+      File in = null;

+      if (document instanceof FileBased)

+      {

+        FileBased fb = (FileBased) document;

+        in = fb.getFile();

+        outName = in.getName() + "_correction_outfile.pdf";

+      }

+      else

+      {

+        in = TempDirHelper.placeInputIntoTempDirFile(document.createInputStream(), "correction_infile.pdf");

+        outName = "correction_outfile.pdf";

+      }

+

+      File out = TempDirHelper.formTempFile(outName);

+

+      String commandline = SettingsReader.getInstance().getSetting(COMMANDLINE_KEY);

+      long timeout = SettingsReader.getInstance().getIntSetting(TIMEOUT_KEY, DEFAULT_TIMEOUT);

+

+      String inF = in.getAbsolutePath();

+      commandline = commandline.replaceFirst(INPUT_DOCUMENT_REPLACE, inF.replaceAll("\\\\", "\\\\\\\\"));

+      String outF = out.getAbsolutePath();

+      commandline = commandline.replaceFirst(OUTPUT_DOCUMENT_REPLACE, outF.replaceAll("\\\\", "\\\\\\\\"));

+

+      log.info(commandline);

+

+      Process p = Runtime.getRuntime().exec(commandline);

+

+      Thread outT = null;

+      Thread errT = null;

+      TimeoutThread tt = null;

+      BufferedReader outReader = null;

+      BufferedReader errReader = null;

+

+      try

+      {

+        outReader = new BufferedReader(new InputStreamReader(p.getInputStream()));

+        errReader = new BufferedReader(new InputStreamReader(p.getErrorStream()));

+

+        outT = new Thread(new ReaderPrinter(outReader, "STDOUT"));

+        errT = new Thread(new ReaderPrinter(errReader, "STDERR"));

+

+        tt = new TimeoutThread(p, timeout, new Thread[] { outT, errT });

+

+        tt.start();

+        outT.start();

+        errT.start();

+

+        log.trace("Joining the STDOUT thread...");

+        outT.join();

+        log.trace("STDOUT thread joined.");

+        log.trace("Joining the STDERR thread...");

+        errT.join();

+        log.trace("STDERR thread joined.");

+

+        log.trace("Waiting for process to end...");

+        p.waitFor();

+        log.trace("process has ended.");

+

+        log.trace("Interrupting timeout thread...");

+        tt.interrupt();

+        log.trace("timeout thread has been interrupted.");

+

+        int exitValue = p.exitValue();

+        log.info("External Corrector exited with: " + exitValue);

+        

+        if (tt.isTimedOut())

+        {

+          throw new CorrectorException(ErrorCode.EXTERNAL_CORRECTOR_TIMEOUT_REACHED, "The external corrector process timed out. timeout = " + timeout);

+        }

+

+        PdfDataSource ds = new FileBasedPdfDataSourceImpl(out, (int) out.length());

+        return ds;

+      }

+      finally

+      {

+        if (outT != null)

+        {

+          outT.interrupt();

+        }

+        if (errT != null)

+        {

+          errT.interrupt();

+        }

+        if (tt != null)

+        {

+          tt.interrupt();

+        }

+        if (outReader != null)

+        {

+          outReader.close();

+        }

+        if (errReader != null)

+        {

+          errReader.close();

+        }

+      }

+

+    }

+    catch (IOException e)

+    {

+      throw new CorrectorException(ErrorCode.CORRECTOR_EXCEPTION, e);

+    }

+    catch (InterruptedException e)

+    {

+      throw new CorrectorException(ErrorCode.CORRECTOR_EXCEPTION, e);

+    }

+    catch (SettingNotFoundException e)

+    {

+      throw new CorrectorException(ErrorCode.CORRECTOR_EXCEPTION, e);

+    }

+    catch (SettingsException e)

+    {

+      throw new CorrectorException(ErrorCode.CORRECTOR_EXCEPTION, e);

+    }

+  }

+

+  protected static class ReaderPrinter implements Runnable

+  {

+    protected BufferedReader reader = null;

+

+    protected String streamName = null;

+

+    public ReaderPrinter(BufferedReader reader, String streamName)

+    {

+      this.reader = reader;

+      this.streamName = streamName;

+    }

+

+    public void run()

+    {

+      try

+      {

+        String line = null;

+

+        while ((line = this.reader.readLine()) != null)

+        {

+          if (line != null)

+          {

+            log.info(streamName + ": " + line);

+          }

+        }

+      }

+      catch (IOException e)

+      {

+        log.error(e);

+      }

+    }

+  }

+

+  protected static class TimeoutThread extends Thread

+  {

+    protected Process proc = null;

+

+    protected long timeout = -1;

+

+    protected boolean ranIntoTimeout = false;

+

+    protected Thread[] threads;

+

+    protected BufferedReader errReader;

+

+    public TimeoutThread(Process proc, long timeout, Thread[] threadsToInterrupt)

+    {

+      this.proc = proc;

+      this.timeout = timeout;

+      this.threads = threadsToInterrupt;

+    }

+

+    public void run()

+    {

+      try

+      {

+        Thread.sleep(this.timeout);

+        log.info("The timeout was reached. Destroying the process.");

+        proc.destroy();

+        ranIntoTimeout = true;

+        log.trace("destroy has been called.");

+        log.trace("Interrupting threads...");

+        for (int i = 0; i < this.threads.length; i++)

+        {

+          this.threads[i].interrupt();

+        }

+        log.trace("threads have been interrupted.");

+      }

+      catch (InterruptedException e)

+      {

+        log.debug("Timeout thread interrupted. This means that the process finished successfully.");

+      }

+    }

+

+    /**

+     * Tells, if the process ran into the timeout.

+     * 

+     * @return Returns true if the timeout was reached. Returns false if the

+     *         timeout was not reached.

+     */

+    public boolean isTimedOut()

+    {

+      return this.ranIntoTimeout;

+    }

+  }

+

+}

diff --git a/src/main/java/at/gv/egiz/pdfas/impl/input/correction/InternalCorrector.java b/src/main/java/at/gv/egiz/pdfas/impl/input/correction/InternalCorrector.java
new file mode 100644
index 0000000..409a600
--- /dev/null
+++ b/src/main/java/at/gv/egiz/pdfas/impl/input/correction/InternalCorrector.java
@@ -0,0 +1,57 @@
+/**

+ * 

+ */

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

+

+import java.io.ByteArrayOutputStream;

+import java.io.IOException;

+

+import at.gv.egiz.pdfas.exceptions.ErrorCode;

+import at.gv.egiz.pdfas.exceptions.framework.CorrectorException;

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

+import at.gv.egiz.pdfas.framework.input.correction.Corrector;

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

+

+import com.lowagie.text.DocumentException;

+import com.lowagie.text.pdf.PdfReader;

+import com.lowagie.text.pdf.PdfStamper;

+

+/**

+ * Corrects a document using iText.

+ * 

+ * @author wprinz

+ */

+public class InternalCorrector implements Corrector

+{

+

+  /**

+   * @see at.gv.egiz.pdfas.framework.input.correction.Corrector#correctDocument(at.gv.egiz.pdfas.framework.input.PdfDataSource)

+   */

+  public PdfDataSource correctDocument(PdfDataSource document) throws CorrectorException

+  {

+    try

+    {

+      byte[] pdf = document.getAsByteArray();

+      PdfReader reader = new PdfReader(pdf);

+

+      ByteArrayOutputStream baos = new ByteArrayOutputStream(pdf.length);

+

+      PdfStamper stamper = new PdfStamper(reader, baos, '\0', false);

+      stamper.close();

+

+      baos.close();

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

+

+      return new ByteArrayPdfDataSourceImpl(corrected_pdf);

+    }

+    catch (DocumentException e)

+    {

+      throw new CorrectorException(ErrorCode.CORRECTOR_EXCEPTION, e);

+    }

+    catch (IOException e)

+    {

+      throw new CorrectorException(ErrorCode.CORRECTOR_EXCEPTION, e);

+    }

+  }

+

+}

diff --git a/src/main/java/at/gv/egiz/pdfas/web/helper/TempDirHelper.java b/src/main/java/at/gv/egiz/pdfas/web/helper/TempDirHelper.java
index 3775776..1b7971c 100644
--- a/src/main/java/at/gv/egiz/pdfas/web/helper/TempDirHelper.java
+++ b/src/main/java/at/gv/egiz/pdfas/web/helper/TempDirHelper.java
@@ -137,7 +137,7 @@ public class TempDirHelper
     return pdfDataSource;

   }

 

-  protected static File placeInputIntoTempDirFile(InputStream input, String fileNameSuffix) throws IOException

+  public static File placeInputIntoTempDirFile(InputStream input, String fileNameSuffix) throws IOException

   {

     String fileName = formatFileName(fileNameSuffix);

 

@@ -156,6 +156,23 @@ public class TempDirHelper
 

     return tmpFile;

   }

+  

+  public static File formTempFile(String fileNameSuffix)

+  {

+    String fileName = formatFileName(fileNameSuffix);

+    File tmpFile = getFileInTempDir(fileName);

+

+    return tmpFile;

+  }

+  

+  protected static File getFileInTempDir (String fileName)

+  {

+    File tempDir = new File(new File(SettingsReader.RESOURCES_PATH), "pdfastmp");

+

+    File tmpFile = new File(tempDir, fileName);

+

+    return tmpFile;

+  }

 

   protected static String formatFileName(String fileNameSuffix)

   {

@@ -176,9 +193,7 @@ public class TempDirHelper
 

   protected static File createTempFileInDir(String fileName) throws IOException

   {

-    File tempDir = new File(new File(SettingsReader.RESOURCES_PATH), "pdfastmp");

-

-    File tmpFile = new File(tempDir, fileName);

+    File tmpFile = getFileInTempDir(fileName);

 

     tmpFile.createNewFile();