Create POI File System from File (#391)

joniles · Nov 18, 2022 · 308e09c · 308e09c
1 parent abd71ea
commit 308e09c
Show file tree

Hide file tree

Showing 3 changed files with 84 additions and 16 deletions.
diff --git a/src/changes/changes.xml b/src/changes/changes.xml
@@ -7,6 +7,7 @@
 	<body>
 		<release date="unreleased" version="10.13.1">
 			<action dev="joniles" type="update">Handle missing default calendar when reading a PMXML file.</action>
+			<action dev="joniles" type="update">When reading an MPP file using a file name or `File` instance, ensure a more memory-efficient approach is used.</action>
 		</release>
 		<release date="2022-11-16" version="10.13.0">
 			<action dev="joniles" type="update">Add support for reading a resource assignment's cost account from P6 schedules.</action>

diff --git a/src/main/java/net/sf/mpxj/mpp/MPPReader.java b/src/main/java/net/sf/mpxj/mpp/MPPReader.java
@@ -23,6 +23,8 @@
 
 package net.sf.mpxj.mpp;
 
+import java.io.File;
+import java.io.FileInputStream;
 import java.io.IOException;
 import java.io.InputStream;
 import java.util.ArrayList;
@@ -35,6 +37,7 @@
 import net.sf.mpxj.ProjectCalendar;
 import net.sf.mpxj.Resource;
 import net.sf.mpxj.TaskField;
+import net.sf.mpxj.common.AutoCloseableHelper;
 import org.apache.poi.poifs.filesystem.DirectoryEntry;
 import org.apache.poi.poifs.filesystem.DocumentEntry;
 import org.apache.poi.poifs.filesystem.DocumentInputStream;
@@ -58,19 +61,12 @@ public final class MPPReader extends AbstractProjectStreamReader
    {
       try
       {
-         //
-         // Open the file system
-         //
-         POIFSFileSystem fs = new POIFSFileSystem(is);
-
-         return read(fs);
-
+         return read(new POIFSFileSystem(is));
       }
+
       catch (IOException ex)
       {
-
          throw new MPXJException(MPXJException.READ_ERROR, ex);
-
       }
    }
 
@@ -79,6 +75,32 @@ public final class MPPReader extends AbstractProjectStreamReader
       return Collections.singletonList(read(inputStream));
    }
 
+   @Override public ProjectFile read(File file) throws MPXJException
+   {
+      POIFSFileSystem fs = null;
+
+      try
+      {
+         // Note we provide this version of the read method rather than using
+         // the AbstractProjectStreamReader version as we can work with the File
+         // instance directly for reduced memory consumption and the ability
+         // to open larger MPP files.
+         fs = new POIFSFileSystem(file);
+         ProjectFile projectFile = read(fs);
+         return projectFile;
+      }
+
+      catch (IOException ex)
+      {
+         throw new MPXJException(MPXJException.READ_ERROR, ex);
+      }
+
+      finally
+      {
+         AutoCloseableHelper.closeQuietly(fs);
+      }
+   }
+
    /**
     * This method allows us to peek into the OLE compound document to extract the file format.
     * This allows the UniversalProjectReader to determine if this is an MPP file, or if

diff --git a/src/main/java/net/sf/mpxj/reader/UniversalProjectReader.java b/src/main/java/net/sf/mpxj/reader/UniversalProjectReader.java
@@ -178,7 +178,30 @@ private List<ProjectFile> readInternal(File file) throws MPXJException
       return readInternal(inputStream);
    }
 
+   /**
+    * Internal implementation of the read method using an InputStream.
+    *
+    * @param inputStream input stream to read from
+    * @return list of schedules
+    */
    private List<ProjectFile> readInternal(InputStream inputStream) throws MPXJException
+   {
+      return readInternal(null, inputStream);
+   }
+
+   /**
+    * Internal implementation of the read method. This primarily works with the
+    * supplied InputStream instance however if the stream is from a file,
+    * we'll pass a File instance too. This allows us to read MPP files
+    * using a more memory-efficient approach based on the File than would
+    * otherwise be possible using an input stream. This also avoids a hard
+    * limit imposed by POI when reading certain very large files.
+    *
+    * @param file optional File instance
+    * @param inputStream input stream to read from
+    * @return list of schedules
+    */
+   private List<ProjectFile> readInternal(File file, InputStream inputStream) throws MPXJException
    {
       try
       {
@@ -226,7 +249,7 @@ private List<ProjectFile> readInternal(InputStream inputStream) throws MPXJExcep
 
          if (matchesFingerprint(buffer, OLE_COMPOUND_DOC_FINGERPRINT))
          {
-            return handleOleCompoundDocument(bis);
+            return handleOleCompoundDocument(file, bis);
          }
 
          if (matchesFingerprint(buffer, MSPDI_FINGERPRINT_1) || matchesFingerprint(buffer, MSPDI_FINGERPRINT_2) || matchesFingerprint(buffer, MSPDI_FINGERPRINT_3))
@@ -405,32 +428,54 @@ private List<ProjectFile> readProjectFile(ProjectReader reader, File file) throw
    /**
     * We have an OLE compound document... but is it an MPP file?
     *
+    * @param file File object
     * @param stream file input stream
     * @return ProjectFile instance
     */
-   private List<ProjectFile> handleOleCompoundDocument(InputStream stream) throws Exception
+   private List<ProjectFile> handleOleCompoundDocument(File file, InputStream stream) throws Exception
    {
       POIFSFileSystem fs;
+      boolean closeFile = false;
 
       try
       {
-         fs = new POIFSFileSystem(new CloseIgnoringInputStream(stream));
+         if (file == null)
+         {
+            fs = new POIFSFileSystem(new CloseIgnoringInputStream(stream));
+         }
+         else
+         {
+            fs = new POIFSFileSystem(file);
+            closeFile = true;
+         }
       }
 
       catch (Exception ex)
       {
          return Collections.emptyList();
       }
 
-      String fileFormat = MPPReader.getFileFormat(fs);
-      if (fileFormat != null && fileFormat.startsWith("MSProject"))
+      try
       {
+         String fileFormat = MPPReader.getFileFormat(fs);
+         if (fileFormat == null || !fileFormat.startsWith("MSProject"))
+         {
+            return Collections.emptyList();
+         }
+
          MPPReader reader = new MPPReader();
          addListenersToReader(reader);
          reader.setProperties(m_properties);
          return Collections.singletonList(reader.read(fs));
       }
-      return Collections.emptyList();
+
+      finally
+      {
+         if (closeFile)
+         {
+            AutoCloseableHelper.closeQuietly(fs);
+         }
+      }
    }
 
    /**
@@ -562,7 +607,7 @@ private List<ProjectFile> handleFile(File file) throws Exception
       try
       {
          fis = new FileInputStream(file);
-         List<ProjectFile> projectFiles = readInternal(fis);
+         List<ProjectFile> projectFiles = readInternal(file, fis);
          fis.close();
          return projectFiles;
       }