XSDVisualizer/XSDVisualiser/Utils/XmlValidation.cs

using System.Xml;
using System.Xml.Schema;

namespace XSDVisualiser.Core;

/// <summary>
/// Validates an XML document against a compiled XSD schema set and a specific global element (node).
/// </summary>
public static class XmlValidator
{
    public static XmlValidationResult ValidateAgainstElement(string xsdPath, string elementName, string? elementNamespace, string xmlPath)
    {
        var set = BuildSchemaSet(xsdPath);
        return ValidateAgainstElement(set, elementName, elementNamespace, xmlPath);
    }

    public static XmlValidationResult ValidateAgainstElement(XmlSchemaSet schemas, string elementName, string? elementNamespace, string xmlPath)
    {
        var result = new XmlValidationResult();

        // Ensure the requested element exists in the schema set
        var qname = new XmlQualifiedName(elementName, elementNamespace ?? string.Empty);
        if (schemas.GlobalElements[qname] is not XmlSchemaElement)
        {
            result.AddError($"Element '{qname}' was not found in the compiled schema set.");
            return result;
        }

        // Probe XML root element
        (string localName, string nsUri)? rootInfo = TryReadRoot(xmlPath);
        if (rootInfo is null)
        {
            result.AddError("Unable to read XML root element.");
            return result;
        }

        var (rootLocal, rootNs) = rootInfo.Value;
        var matchesRoot = string.Equals(rootLocal, elementName, StringComparison.Ordinal) && string.Equals(rootNs ?? string.Empty, elementNamespace ?? string.Empty, StringComparison.Ordinal);

        var settings = new XmlReaderSettings
        {
            DtdProcessing = DtdProcessing.Ignore,
            ValidationType = ValidationType.Schema,
            Schemas = schemas,
            CloseInput = true,
            ConformanceLevel = ConformanceLevel.Auto
        };
        settings.ValidationFlags = XmlSchemaValidationFlags.ReportValidationWarnings | XmlSchemaValidationFlags.ProcessIdentityConstraints;

        void Handler(object? sender, ValidationEventArgs e)
        {
            if (e.Severity == XmlSeverityType.Warning)
                result.AddWarning(e.Message, e.Exception?.LineNumber, e.Exception?.LinePosition);
            else
                result.AddError(e.Message, e.Exception?.LineNumber, e.Exception?.LinePosition);
        }

        settings.ValidationEventHandler += Handler;

        if (matchesRoot)
        {
            using var reader = XmlReader.Create(xmlPath, settings);
            try
            {
                while (reader.Read())
                {
                    // just advance to trigger validation callbacks
                }
            }
            catch (XmlException xe)
            {
                result.AddError($"XML parsing error: {xe.Message}", xe.LineNumber, xe.LinePosition);
            }
            return result;
        }
        else
        {
            // Root does not match the selected schema element. Try to locate the first matching subtree and validate only that fragment.
            // This enables validating an XML file towards a selected node from the XSD.
            var fragmentSettings = new XmlReaderSettings
            {
                DtdProcessing = DtdProcessing.Ignore,
                ValidationType = ValidationType.Schema,
                Schemas = schemas,
                CloseInput = true,
                ConformanceLevel = ConformanceLevel.Fragment
            };
            fragmentSettings.ValidationFlags = settings.ValidationFlags;
            fragmentSettings.ValidationEventHandler += Handler;

            try
            {
                var (elementNode, loadError) = FindFirstElementNode(xmlPath, elementName, elementNamespace);
                if (loadError is not null)
                {
                    result.AddError(loadError.Value.Message, loadError.Value.LineNumber, loadError.Value.LinePosition);
                    return result;
                }

                if (elementNode is null)
                {
                    result.AddError($"Could not find any element '{{{elementNamespace}}}{elementName}' in the XML document to validate against.");
                    return result;
                }

                // Inform as a warning that we validate a subtree instead of the document root
                result.AddWarning($"Validating against the first occurrence of '{{{elementNamespace}}}{elementName}' found in the document (root is '{{{rootNs}}}{rootLocal}').");

                using var nodeReader = new XmlNodeReader(elementNode);
                using var validatingReader = XmlReader.Create(nodeReader, fragmentSettings);
                while (validatingReader.Read())
                {
                    // advance to trigger validation callbacks for the subtree
                }
            }
            catch (XmlException xe)
            {
                result.AddError($"XML parsing error: {xe.Message}", xe.LineNumber, xe.LinePosition);
            }

            return result;
        }
    }

    private static XmlSchemaSet BuildSchemaSet(string xsdPath)
    {
        var set = new XmlSchemaSet
        {
            XmlResolver = new XmlUrlResolver(),
            CompilationSettings = new XmlSchemaCompilationSettings { EnableUpaCheck = true }
        };

        using var reader = XmlReader.Create(xsdPath, new XmlReaderSettings { DtdProcessing = DtdProcessing.Ignore });
        var schema = XmlSchema.Read(reader, null);
        if (schema != null)
            set.Add(schema);
        set.Compile();
        return set;
    }

    private static (string localName, string nsUri)? TryReadRoot(string xmlPath)
    {
        using var reader = XmlReader.Create(xmlPath, new XmlReaderSettings { DtdProcessing = DtdProcessing.Ignore, IgnoreWhitespace = true, IgnoreComments = true });
        try
        {
            while (reader.Read())
            {
                if (reader.NodeType == XmlNodeType.Element && reader.Depth == 0)
                {
                    return (reader.LocalName, reader.NamespaceURI);
                }
            }
        }
        catch
        {
            // ignored; higher level will report XmlException separately
        }
        return null;
    }

    private static (XmlElement? Node, (string Message, int LineNumber, int LinePosition)? LoadError) FindFirstElementNode(string xmlPath, string elementName, string? elementNamespace)
    {
        try
        {
            var xr = XmlReader.Create(xmlPath, new XmlReaderSettings { DtdProcessing = DtdProcessing.Ignore });
            var doc = new XmlDocument();
            doc.PreserveWhitespace = false;
            doc.Load(xr);

            static XmlElement? Traverse(XmlNode node, string name, string? ns)
            {
                if (node is XmlElement el)
                {
                    if (string.Equals(el.LocalName, name, StringComparison.Ordinal) && string.Equals(el.NamespaceURI ?? string.Empty, ns ?? string.Empty, StringComparison.Ordinal))
                        return el;
                }
                foreach (XmlNode child in node.ChildNodes)
                {
                    var found = Traverse(child, name, ns);
                    if (found != null) return found;
                }
                return null;
            }

            var match = Traverse(doc, elementName, elementNamespace);
            return (match, null);
        }
        catch (XmlException xe)
        {
            return (null, ($"XML parsing error: {xe.Message}", xe.LineNumber, xe.LinePosition));
        }
    }
}

public sealed class XmlValidationResult
{
    private readonly List<XmlValidationIssue> _issues = new();

    public bool IsValid => _issues.TrueForAll(i => i.Severity != XmlSeverityType.Error);

    public IReadOnlyList<XmlValidationIssue> Issues => _issues;

    public IEnumerable<XmlValidationIssue> Errors => _issues.Where(i => i.Severity == XmlSeverityType.Error);
    public IEnumerable<XmlValidationIssue> Warnings => _issues.Where(i => i.Severity == XmlSeverityType.Warning);

    internal void AddError(string message, int? line = null, int? position = null) =>
        _issues.Add(new XmlValidationIssue(XmlSeverityType.Error, message, line ?? 0, position ?? 0));

    internal void AddWarning(string message, int? line = null, int? position = null) =>
        _issues.Add(new XmlValidationIssue(XmlSeverityType.Warning, message, line ?? 0, position ?? 0));
}

public sealed record XmlValidationIssue(XmlSeverityType Severity, string Message, int LineNumber, int LinePosition);