Why can't this file be deleted after using C1ZipFile?

Question

The following code gives me a System.IO.IOException with the message 'The process cannot access the file'.

private void UnPackLegacyStats()
{
  DirectoryInfo oDirectory;
  XmlDocument oStatsXml;

  //Get the directory
  oDirectory = new DirectoryInfo(msLegacyStatZipsPath);

  //Check if the directory exists
  if (oDirectory.Exists)
  {
    //Loop files
    foreach (FileInfo oFile in oDirectory.GetFiles())
    {
      //Check if file is a zip file
      if (C1ZipFile.IsZipFile(oFile.FullName))
      {
        //Open the zip file
        using (C1ZipFile oZipFile = new C1ZipFile(oFile.FullName, false))
        {
          //Check if the zip contains the stats
          if (oZipFile.Entries.Contains("Stats.xml"))
          {
            //Get the stats as a stream
            using (Stream oStatsStream = oZipFile.Entries["Stats.xml"].OpenReader())
            {
              //Load the stats as xml
              oStatsXml = new XmlDocument();
              oStatsXml.Load(oStatsStream);

              //Close the stream
              oStatsStream.Close();
            }

            //Loop hit elements
            foreach (XmlElement oHitElement in oStatsXml.SelectNodes("/*/hits"))
            {
              //Do stuff
            }                
          }

          //Close the file
          oZipFile.Close();
        }
      }

      //Delete the file
      oFile.Delete();
    }
  }
}

I am struggling to see where the file could still be locked. All objects that could be holding onto a handle to the file are in using blocks and are explicitly closed.

Is it something to do with using FileInfo objects rather than the strings returned by the static GetFiles method?

Any ideas?

Answer 1

I'm just guessing: are you sure that oZipFile.Close() is enough? Perhaps you have to call oZipFile.Dispose() or oZipFile.Finalize() to be sure it has actually released the resources.

Answer 2

I do not see problems in your code, everything look ok. To check is the problem lies in C1ZipFile I suggest you initialize zip from stream, instead of initialization from file, so you close stream explicitly:

//Open the zip file
using (Stream ZipStream = oFile.OpenRead())
using (C1ZipFile oZipFile = new C1ZipFile(ZipStream, false))
{
    // ...

Several other suggestions:

• You do not need to call Close() method, with using (...), remove them.
• Move xml processing (Loop hit elements) outsize zip processing, i.e. after zip file closeing, so you keep file opened as least as possible.

Answer 3

I assume you're getting the error on the oFile.Delete call. I was able to reproduce this error. Interestingly, the error only occurs when the file is not a zip file. Is this the behavior you are seeing?

It appears that the C1ZipFile.IsZipFile call is not releasing the file when it's not a zip file. I was able to avoid this problem by using a FileStream instead of passing the file path as a string (the IsZipFile function accepts either).

So the following modification to your code seems to work:

if (oDirectory.Exists)
{
    //Loop files
    foreach (FileInfo oFile in oDirectory.GetFiles())
    {
        using (FileStream oStream = new FileStream(oFile.FullName, FileMode.Open))
        {
            //Check if file is a zip file
            if (C1ZipFile.IsZipFile(oStream))
            {
            // ...
            }
        }
        //Delete the file
        oFile.Delete();
    }
}

In response to the original question in the subject: I don't know if it's possible to know if a file can be deleted without attempting to delete it. You could always write a function that attempts to delete the file and catches the error if it can't and then returns a boolean indicating whether the delete was successful.

Answer 4

More then Likely it's not being disposed, anytime you access something outside of managed code(streams, files, etc.) you MUST dispose of them. I learned the hard way with Asp.NET and Image files, it will fill up your memory, crash your server, etc.

Answer 5

In the interest of completeness I am posing my working code as the changes came from more than one source.

private void UnPackLegacyStats()
{
  DirectoryInfo oDirectory;
  XmlDocument oStatsXml;

  //Get the directory
  oDirectory = new DirectoryInfo(msLegacyStatZipsPath);

  //Check if the directory exists
  if (oDirectory.Exists)
  {
    //Loop files
    foreach (FileInfo oFile in oDirectory.GetFiles())
    {
      //Set empty xml
      oStatsXml = null;

      //Load file into a stream
      using (Stream oFileStream = oFile.OpenRead())
      {
        //Check if file is a zip file
        if (C1ZipFile.IsZipFile(oFileStream))
        {
          //Open the zip file
          using (C1ZipFile oZipFile = new C1ZipFile(oFileStream, false))
          {
            //Check if the zip contains the stats
            if (oZipFile.Entries.Contains("Stats.xml"))
            {
              //Get the stats as a stream
              using (Stream oStatsStream = oZipFile.Entries["Stats.xml"].OpenReader())
              {
                //Load the stats as xml
                oStatsXml = new XmlDocument();
                oStatsXml.Load(oStatsStream);
              }
            }
          }
        }
      }

      //Check if we have stats
      if (oStatsXml != null)
      {
        //Process XML here
      }

      //Delete the file
      oFile.Delete();
    }
  }
}

The main lesson I learned from this is to manage file access in one place in the calling code rather than letting other components manage their own file access. This is most apropriate when you want to use the file again after the other component has finished it's task.

Although this takes a little more code you can clearly see where the stream is disposed (at the end of the using), compared to having to trust that a component has correctly disposed of the stream.