Using XML CDATA nodes to send files via a Web Service
Using XML CDATA nodes to send files via a web service.
Introduction
There was a need to have a web service send a file. The documented way to do this is through WSE (Web Service Extension), which is an add-on to the .NET Framework to do things with web services like attachments. Using two CodeProject articles, I added zipping the file and encoding the file using Base64. Then I placed the encoded file into a CDATA node in an XML document.
Background
As I looked into sending a file via a web service, I was not pleased with the amount of work that had to be done to attach a file using WSE. So I started to look for other options. One of my co-workers suggested using a CDATA node in an XML document. As I looked into the CDATA node, it seemed like it might be a good fit. The CDATA node is for storing blob data in an XML file. It is marked so that it ignores what is between the CDATA nodes. I decided to zip the file to reduce the size of the XML file I would be sending. The problem is that zipping a file produces some characters that cause problems with the XML file. So, that leads us to base64 encoding. I found two code project articles to help me with zipping files with Sharpziplib and base64 encoding that would fit into what I needed.
The code
After having some discussion with my co-workers, we decided on an XML structure that looks like this:
<FILES>
<FILE FileName="file.txt"><CDATA></CDATA></File>
<FILE FileName="file2.txt"><CDATA></CDATA></File>
</FILES>
where we could add as many file nodes in the XML document as we wanted. Here is the code to create that XML doc:
public static XmlDocument GetFileXml(String[] Files)
{
//We are returning an XML Doc
XmlDocument tmpXmlDoc = new XmlDocument();
//Create the root node...
tmpXmlDoc.LoadXml("<FILES></FILES>");
//Loop through the WorkItemFiles
//array and create the file nodes
for (Int32 i=0;i<=Files.Length-1;i++)
{
//The File node
XmlElement tmpFile =
tmpXmlDoc.CreateElement("File");
//Add the filename attribute to it...
XmlAttribute tmpA =
tmpXmlDoc.CreateAttribute("FileName");
//Find the last \ to pull off the filename.
tmpA.Value =
Files[i].Substring(Files[i].LastIndexOf(@"\")+1);
tmpFile.Attributes.Append(tmpA);
//Read in the File
Byte[] myCompressed;
//Compress the file
myCompressed =
CompressionHelper.CompressFile(Files[i]);
//Next encode the compressed stream
//so that we don't blow the
//CData node in the XML
Char[] myEncoded;
Base64Encoder myBE;
myBE = new Base64Encoder(myCompressed);
myEncoded = myBE.GetEncoded();
//A StringBuilder is the fastest way
//to get a string from a byte array
StringBuilder myStr = new StringBuilder();
myStr.Append(myEncoded);
//create CData child appened
//to the file node...
tmpFile.AppendChild(tmpXmlDoc.CreateCDataSection(
myStr.ToString()));
//Add the file node to the xml doc...
tmpXmlDoc.DocumentElement.AppendChild(tmpFile);
} // for
return tmpXmlDoc;
}
Next, we call the web service to get the XML document. Then, we parse through the XML file and extract and save the documents from the CDATA nodes.
public static void GetFiles(XmlNode inXml, String FilePath)
{
XmlNodeList myFile;
myFile = inXml.SelectNodes("/File");
//This Loops through the File nodes
for (Int32 j=0;j<=myFile.Count-1;j++)
{
//The first child is the CDATA node
XmlNode myCData = (XmlNode)(myFile.Item(j)).FirstChild;
//Grab the Filename attribute
String FileName =
((XmlNode)(myFile.Item(j))).Attributes.Item(0).InnerText;
//This is the actual data that is in the CDATA
Char[] theFile;
theFile = new Char[myCData.InnerText.Length];
for (Int32 i=0;i<=myCData.InnerText.Length-1;i++)
{
theFile[i] = myCData.InnerText[i];
}//for i
//Delete the file if it exists
if (File.Exists(FilePath+FileName))
{
File.Delete(FilePath+FileName);
}
//next we need to switch from base64
//encoding to our zip compressed
Byte[] myDecoder;
Base64Decoder myBD = new Base64Decoder(theFile);
myDecoder = myBD.GetDecoded();
//Finally we decompress the file and
//save it where it is supose to go
CompressionHelper.DecompressFile(myDecoder,
FilePath+FileName);
} //for j
}
Conclusion
I would like to thank my co-worker Steve Rowe for his help and guidance in coming to this solution. I would also like to thank Uwe Keim for his shareziplib article and wchvic for his base64 article. Hopefully, others will find this article helpful in their solutions.