Reads and writes Ariba CIF files.

• .NET Framework 4.0 (should work with 3.5 with some changes)

Reading basically works along these lines:

using(var streamReader = new StreamReader(...))
using(var reader = new CIFReader(streamReader))
{
   reader.Read(protocol);
}

where protocol is a class that implements the ICIFReaderProtocol interface (you can derive it from the abstract base class CIFReaderProtocol). It basically defines three methods that will be called during reading:

public interface ICIFReaderProtocol
{
    // Called when header is ready
    void HandleHeader(CIFHeader header);

	// Called when a new item is ready
	void HandleItem(CIFItem item);

	// Called after trailer is ready
	void HandleTrailer(CIFTrailer trailer);
}

Write CIF files with code like this:

using(var streamWriter = new StreamWriter(...))
using(var writer = new CIFWriter(streamWriter))
{
  writer.Write(protocol);
}

where, again, protocol implements an interface ICIFWriterProtocol (you can, again, derive from the abstract base class CIFWriterProtocol). Here's ICIFWriterProtocol:

public interface ICIFWriterProtocol
{
    // Return header here
	CIFHeader GetCIFHeader();

	// Return list of fields to export.
	string[] GetCIFFieldNames();

	// Return enumerator for CIFItem instances here
	IEnumerable<CIFItem> GetCIFItems();

	// Return trailer here
	CIFTrailer GetCIFTrailer();
}

1. Reading is basically completed. All green. We're successfully reading and parsing all samples from the spec.
2. There's some glitches in the spec, e.g. they mention Units of Measure in the spec, but use Unit of Measure in the examples. We're currently sticking to the spec (so end up reading Units of Measure). If you want Unit of Measure, use the indexer.
3. Writing should be okay, but it's not as well tested as reading. There also might be a lot of edge cases.

Having said that, I still don't consider this production ready.