TurboXml

TurboXml is a .NET library that provides a lightweight and fast SAX - Simple API XML parser by using callbacks.

This is the equivalent of System.Xml.XmlReader but faster with no allocations. 🚀

✨ Features

• Should be slightly faster than System.Xml.XmlReader
• Zero Allocation XML Parser
  • Callbacks received ReadOnlySpan<char> for the parsed elements.
  • Parse from small to very large XML documents, without allocating!
• Optimized with SIMD
  • TurboXml is using some SIMD to improve parsing of large portions of XML documents.
• Provide precise source location of the XML elements parsed (to report warning/errors)
• Compatible with net8.0+
• NativeAOT ready

📃 User Guide

TurboXML is in the family of the SAX parsers and so you need to implement the callbacks defined by IXmlReadHandler.

By default this handler implements empty interface methods that you can easily override:

var xml = "<?xml version=\"1.0\"?><root enabled=\"true\">Hello World!</root>";
var handler = new MyXmlHandler();
XmlParser.Parse(xml, ref handler);
// Will print:
//
// BeginTag(1:23): root
// Attribute(1:28)-(1:36): enabled="true"
// Content(1:43): Hello World!
// EndTag(1:57): root

struct MyXmlHandler : IXmlReadHandler
{
    public void OnBeginTag(ReadOnlySpan<char> name, int line, int column)
        => Console.WriteLine($"BeginTag({line + 1}:{column + 1}): {name}");

    public void OnEndTagEmpty()
        => Console.WriteLine($"EndTagEmpty");

    public void OnEndTag(ReadOnlySpan<char> name, int line, int column)
        => Console.WriteLine($"EndTag({line + 1}:{column + 1}): {name}");

    public void OnAttribute(ReadOnlySpan<char> name, ReadOnlySpan<char> value, int nameLine, int nameColumn, int valueLine, int valueColumn)
        => Console.WriteLine($"Attribute({nameLine + 1}:{nameColumn + 1})-({valueLine + 1}:{valueColumn + 1}): {name}=\"{value}\"");

    public void OnText(ReadOnlySpan<char> text, int line, int column)
        => Console.WriteLine($"Content({line + 1}:{column + 1}): {text}");
}

📊 Benchmarks

The solution contains 2 benchmarks:

• BenchStream that parses 240+ MSBuild xml files (targets and props) from the .NET 8 (or latest SDK) installed
• BenchString that parses the Tiger.svg in memory from a string.

In general, the advantages of TurboXml over System.Xml.XmlReader:

• It should be slightly faster - from 10% to 30% - or more, specially if tag names, attributes or even content are bigger than 8 consecutive characters by using SIMD instructions.
• It will make almost zero allocations - apart for the internal buffers used to pass data as ReadOnlySpan<char> back the the XML Handler.

Stream Results

BenchmarkDotNet v0.13.12, Windows 11 (10.0.22631.3085/23H2/2023Update/SunValley3)
AMD Ryzen 9 7950X, 1 CPU, 32 logical and 16 physical cores
.NET SDK 8.0.101
  [Host]     : .NET 8.0.1 (8.0.123.58001), X64 RyuJIT AVX-512F+CD+BW+DQ+VL+VBMI
  DefaultJob : .NET 8.0.1 (8.0.123.58001), X64 RyuJIT AVX-512F+CD+BW+DQ+VL+VBMI

String Results

🚨 XML Conformance and Known Limitations

This parser is following the Extensible Markup Language (XML) 1.0 (Fifth Edition) and should support any XML valid documents, except for the known limitations described below:

• For simplicity of the implementation, this parser does not support DTD, custom entities and XML directives (<!DOCTYPE ...>). If you are looking for this, you should instead use System.Xml.XmlReader.
• This parser checks for well formed XML, matching begin and end tags and report an error if they are not matching
• This parser does not check for duplicated attributes.
  • It is the responsibility of the XML handler to implement such a check. The rationale is that the check can be performed more efficiently depending on user scenarios (e.g bit flags...etc.)

🏗️ Build

You need to install the .NET 8 SDK. Then from the root folder:

$ dotnet build src -c Release

🪪 License

This software is released under the BSD-2-Clause license.

🤗 Author

Alexandre Mutel aka xoofx.