BeanSpitter

A Simple, Stream-Oriented XML Parser for .NET - NuGet: here

This XML Parsing Library aims to simplify parsing operations for files which:

• Have an associated XML Schema Definition (XSD);
• Have classes generated from its XSD file (from xsd.exe, for example);
• You don't want to create the parsing code by hand, going through XElement, dealing with strings and etc., especially for complex XML Schemas.

What does this library offers:

• It is built over regular XmlReader and System.Xml infrastructure, so no "out of the world" code.
• It is Multi-Platform: net45 and netstandard2.0
• Stream-oriented, Memory friendly (around 120 MB of Total Memory Consumption while running tests with over 4GB XML files with Visual Studio), asynchronous XML Validation and Parsing (using Stream and Task);
• Schema Loading, XML Validation and Parsing methods with overloads for Byte Arrays, Files and Streams;
• Parsing by type, with header type support (like those XML Files that have Envelope, Header and Payload schemas, e.g. ISO 20022 XML Messages - this library was initially made for these);
• Async Event-raising for nodes read, errors occurred and when it finishes parsing and validating. Kind of SAX inspired, but it raises POCO-like objects;

Usage and Documentation:

• Still a work in progress, but you can find code samples in the SAMPLES folder and a very basic usage example can be found here

Caveats:

• When using .NET Core, you must add a new XmlUrlResolver() to your XmlSchemaSet.XmlResolver property (reference here) or do this.

Other remarks:

• It has only one dependency, for better testability, which is System.IO.Abstractions;
• The idea is not to "hog" your ThreadPool: It will set the TaskScheduler to Environment.ProcessorCount * 2 by default. Reference from here.
• It is modular: Even though the defaults work OK, if you fancy something different, the object constructors from the library get Interfaces for almost everything, so you can replace even the EventRiser implementation if you would like.

Misc

• It is my first "real" opensource project, so bear with me :). Suggestions and contributions are welcome.
• It is not yet a fully tested solution (you can help to improve that), but it is good enough for what I meant and the current tests cover quite complex scenarios out-of-the-box
• To contribute, clone it, create a branch, do your stuff (with tests, please :)) and submit a pull request.