Migrate TagExpressions to source-generated regexes (multi-target netstandard2.0 + net8.0)

src/Schematron/Config.cs

@@ -143,7 +143,7 @@ public static void Setup()
     {
         System.Diagnostics.Trace.Write("Loading schematron statics...");
         System.Diagnostics.Trace.Write(CompiledExpressions.Schema.ReturnType);
-        System.Diagnostics.Trace.Write(TagExpressions.Dir.RightToLeft);
+        System.Diagnostics.Trace.Write(TagExpressions.Dir().RightToLeft);
         System.Diagnostics.Trace.WriteLine(FormattingUtils.XmlErrorPosition.RightToLeft);
     }
 }

src/Schematron/Formatters/LogFormatter.cs

@@ -17,7 +17,7 @@ public override void Format(Test source, XPathNavigator context, StringBuilder o
     {
         var sb = FormatMessage(source, context, source.Message);
         // Finally remove any non-name schematron tag in the message.
-        var res = TagExpressions.AllSchematron.Replace(sb.ToString(), string.Empty);
+        var res = TagExpressions.AllSchematron().Replace(sb.ToString(), string.Empty);
         sb = new StringBuilder();
         if (source is Assert)
         {

src/Schematron/Formatters/XmlFormatter.cs

@@ -41,7 +41,7 @@ public override void Format(Test source, XPathNavigator context, StringBuilder o
         msg = FormatMessage(source, context, msg).ToString();
 
         // Finally remove any non-name schematron tag in the message.
-        var res = TagExpressions.AllSchematron.Replace(msg, string.Empty);
+        var res = TagExpressions.AllSchematron().Replace(msg, string.Empty);
 
         //Accumulate namespaces found during traversal of node for its position.
         var ns = new Hashtable();

src/Schematron/Schematron.csproj

@@ -3,7 +3,7 @@
   <PropertyGroup>
     <AssemblyName>Schematron</AssemblyName>
     <RootNamespace>Schematron</RootNamespace>
-    <TargetFramework>netstandard2.0</TargetFramework>
+    <TargetFrameworks>netstandard2.0;net8.0</TargetFrameworks>
     <GenerateDocumentationFile>true</GenerateDocumentationFile>
     <PackageId>Schematron.NET</PackageId>
     <LangVersion>Latest</LangVersion>
@@ -13,7 +13,7 @@
     <InternalsVisibleTo Include="Schematron.Tests" />
   </ItemGroup>
 
-  <ItemGroup>
+  <ItemGroup Condition="'$(TargetFramework)' == 'netstandard2.0'">
     <PackageReference Include="PolySharp" Version="1.15.0">
       <PrivateAssets>all</PrivateAssets>
       <IncludeAssets>runtime; build; native; contentfiles; analyzers; buildtransitive</IncludeAssets>

src/Schematron/TagExpressions.cs

@@ -3,41 +3,72 @@
 namespace Schematron;
 
 /// <summary />
-class TagExpressions
+partial class TagExpressions
 {
+#if NET8_0_OR_GREATER
     /// <summary>
     /// The compiled regular expression to replace the special <c>name</c> and <c>value</c> tags inside a message.
     /// </summary>
     /// <remarks>
     /// Replaces each instance of <c>name</c> and <c>value</c>tags with the value in the current context element.
     /// </remarks>
-    public static Regex NameValueOf;
-
-    public static Regex Emph;
-    public static Regex Dir;
-    public static Regex Span;
-    public static Regex Para;
-    public static Regex Any;
-    public static Regex AllSchematron;
-
-    /// <summary />
-    static TagExpressions()
-    {
-        // The element declarations can contain the namespace if expanded in a loaded document.
-        NameValueOf = new Regex(@"<[^\s>]*\b(name|value-of)\b[^>]*/>", RegexOptions.Compiled);
-        Emph = new Regex(@"<[^\s>]*\bemph\b[^>]*>", RegexOptions.Compiled);
-        Dir = new Regex(@"<[^\s]*\bdir\b[^>]*>", RegexOptions.Compiled);
-        Span = new Regex(@"<[^\s]*\bspan\b[^>]*>", RegexOptions.Compiled);
-        Para = new Regex(@"<[^\s]*\bp\b[^>]*>", RegexOptions.Compiled);
-        Any = new Regex(@"<[^\s]*[^>]*>", RegexOptions.Compiled);
-        // Closing elements don't have an expanded xmlns so they will be matched too.
-        // TODO: improve this to avoid removing non-schematron closing elements.
-        var nsPattern = "(?:" + Regex.Escape(Schema.LegacyNamespace) + "|" + Regex.Escape(Schema.IsoNamespace) + ")";
-        AllSchematron = new Regex(@"<.*\bxmlns\b[^\s]*" + nsPattern + "[^>]*>|</[^>]*>", RegexOptions.Compiled);
-    }
-
-    TagExpressions()
-    {
-    }
+    // The element declarations can contain the namespace if expanded in a loaded document.
+    [GeneratedRegex(@"<[^\s>]*\b(name|value-of)\b[^>]*/>")]
+    public static partial Regex NameValueOf();
+
+    [GeneratedRegex(@"<[^\s>]*\bemph\b[^>]*>")]
+    public static partial Regex Emph();
+
+    [GeneratedRegex(@"<[^\s]*\bdir\b[^>]*>")]
+    public static partial Regex Dir();
+
+    [GeneratedRegex(@"<[^\s]*\bspan\b[^>]*>")]
+    public static partial Regex Span();
+
+    [GeneratedRegex(@"<[^\s]*\bp\b[^>]*>")]
+    public static partial Regex Para();
+
+    [GeneratedRegex(@"<[^\s]*[^>]*>")]
+    public static partial Regex Any();
+
+    // Closing elements don't have an expanded xmlns so they will be matched too.
+    // TODO: improve this to avoid removing non-schematron closing elements.
+    // Pattern derived from Regex.Escape(Schema.LegacyNamespace) and Regex.Escape(Schema.IsoNamespace).
+    // If those constants ever change, update the hardcoded escaped values here to match.
+    [GeneratedRegex(@"<.*\bxmlns\b[^\s]*(?:http://www\.ascc\.net/xml/schematron|http://purl\.oclc\.org/dsdl/schematron)[^>]*>|</[^>]*>")]
+    public static partial Regex AllSchematron();
+#else
+    /// <summary>
+    /// The compiled regular expression to replace the special <c>name</c> and <c>value</c> tags inside a message.
+    /// </summary>
+    /// <remarks>
+    /// Replaces each instance of <c>name</c> and <c>value</c>tags with the value in the current context element.
+    /// </remarks>
+    // The element declarations can contain the namespace if expanded in a loaded document.
+    static readonly Regex _nameValueOf = new Regex(@"<[^\s>]*\b(name|value-of)\b[^>]*/>", RegexOptions.Compiled);
+    public static Regex NameValueOf() => _nameValueOf;
+
+    static readonly Regex _emph = new Regex(@"<[^\s>]*\bemph\b[^>]*>", RegexOptions.Compiled);
+    public static Regex Emph() => _emph;
+
+    static readonly Regex _dir = new Regex(@"<[^\s]*\bdir\b[^>]*>", RegexOptions.Compiled);
+    public static Regex Dir() => _dir;
+
+    static readonly Regex _span = new Regex(@"<[^\s]*\bspan\b[^>]*>", RegexOptions.Compiled);
+    public static Regex Span() => _span;
+
+    static readonly Regex _para = new Regex(@"<[^\s]*\bp\b[^>]*>", RegexOptions.Compiled);
+    public static Regex Para() => _para;
+
+    static readonly Regex _any = new Regex(@"<[^\s]*[^>]*>", RegexOptions.Compiled);
+    public static Regex Any() => _any;
+
+    // Closing elements don't have an expanded xmlns so they will be matched too.
+    // TODO: improve this to avoid removing non-schematron closing elements.
+    static readonly Regex _allSchematron = new Regex(
+        @"<.*\bxmlns\b[^\s]*(?:" + Regex.Escape(Schema.LegacyNamespace) + "|" + Regex.Escape(Schema.IsoNamespace) + @")[^>]*>|</[^>]*>",
+        RegexOptions.Compiled);
+    public static Regex AllSchematron() => _allSchematron;
+#endif
 }
 

src/Schematron/Test.cs

@@ -22,7 +22,7 @@ public Test(string test, string message) : base(test)
         // Save <name> and <value-of> tags in the message and their paths / selects in their compiled form.
         // TODO: see if we can work with the XML in the message, instead of using RE.
         // TODO: Check the correct usage of path and select attributes.
-        NameValueOfExpressions = TagExpressions.NameValueOf.Matches(Message);
+        NameValueOfExpressions = TagExpressions.NameValueOf().Matches(Message);
         var nc = NameValueOfExpressions.Count;
         NamePaths = new XPathExpression[nc];
         ValueOfSelects = new XPathExpression[nc];