Repository metrics
- Stars
- (20,414 stars)
- PR merge metrics
- (Avg merge 6d 17h) (256 merged PRs in 30d)
Description
Version Used: 3.0.0
Consider the following code:
using System;
using System.Linq;
using Microsoft.CodeAnalysis;
using Microsoft.CodeAnalysis.CSharp;
using Microsoft.CodeAnalysis.CSharp.Syntax;
class Program
{
static void Main()
{
string code = @"
class C
{
static void Main()
{
_ = 0;
}
}";
var file = SyntaxFactory.ParseCompilationUnit(code);
void Compile()
{
var tree = SyntaxFactory.SyntaxTree(file);
var compilation = CSharpCompilation.Create(null)
.AddSyntaxTrees(tree)
.AddReferences(MetadataReference.CreateFromFile(typeof(object).Assembly.Location));
var diagnostics = compilation.GetDiagnostics();
if (!diagnostics.Any())
Console.WriteLine("No diagnostics");
foreach (var diagnostic in diagnostics)
{
Console.WriteLine(diagnostic);
}
}
Compile();
var identifier = file.DescendantNodes().OfType<IdentifierNameSyntax>().Single();
file = file.ReplaceNode(identifier, SyntaxFactory.IdentifierName("_"));
Compile();
}
}
It parses a simple piece of code and compiles it. Then, it replaces the parsed identifier _ with one created using SyntaxFactory.IdentifierName("_") and compiles again.
I would expect both to compile by interpreting the _ as a discard, but they don't; the output of the program above is:
No diagnostics
(6,1): error CS0103: The name '_' does not exist in the current context
As far as I can tell, this happens because the identifier token created by SyntaxFactory does not have its contextual kind set up correctly:
var fromFactory = SyntaxFactory.Identifier("_");
var parsed = ((IdentifierNameSyntax)SyntaxFactory.ParseExpression("_")).Identifier;
SyntaxKind ContextualKind(SyntaxToken token)
{
var property = typeof(SyntaxToken).GetProperty("RawContextualKind", BindingFlags.NonPublic | BindingFlags.Instance);
return (SyntaxKind)(int)property.GetValue(token);
}
Console.WriteLine(ContextualKind(fromFactory));
Console.WriteLine(ContextualKind(parsed));
This prints:
IdentifierToken
UnderscoreToken
There is an overload of SyntaxFactory.Identifier which accepts SyntaxKind contextualKind, so it is possible to use SyntaxFactory to create the right identifier token. But shouldn't the simpler overload do this automatically?