Search icon CANCEL
Arrow left icon
Explore Products
Best Sellers
New Releases
Books
Videos
Audiobooks
Learning Hub
Conferences
Free Learning
Arrow right icon
Arrow up icon
GO TO TOP
Roslyn Cookbook

You're reading from   Roslyn Cookbook Compiler as a Service, Code Analysis, Code Quality and more

Arrow left icon
Product type Paperback
Published in Jul 2017
Publisher Packt
ISBN-13 9781787286832
Length 350 pages
Edition 1st Edition
Languages
Arrow right icon
Author (1):
Arrow left icon
Manish Vasani Manish Vasani
Author Profile Icon Manish Vasani
Manish Vasani
Arrow right icon
View More author details
Toc

Table of Contents (11) Chapters Close

Preface 1. Writing Diagnostic Analyzers FREE CHAPTER 2. Consuming Diagnostic Analyzers in .NET Projects 3. Writing IDE Code Fixes, Refactorings, and Intellisense Completion Providers 4. Improving Code Maintenance of C# Code Base 5. Catch Security Vulnerabilities and Performance Issues in C# Code 6. Live Unit Testing in Visual Studio Enterprise 7. C# Interactive and Scripting 8. Contribute Simple Functionality to Roslyn C# Compiler Open Source Code 9. Design and Implement a New C# Language Feature 10. Command-Line Tools Based on Roslyn API

Creating a symbol analyzer to report issues about symbol declarations

A symbol analyzer registers action callbacks to analyze one or more kinds of symbol declarations, such as types, methods, fields, properties, events, and so on, reports semantic issues about declarations.

In this section, we will create a symbol analyzer that extends the compiler diagnostic CS0542 (member names cannot be the same as their enclosing type) to report a diagnostic if member names are the same as any of the outer parent type. For example, the analyzer will report a diagnostic for the innermost type NestedClass here:

public class NestedClass
{
public class InnerClass
{
public class NestedClass
{
}
}
}

Getting ready

You will need to have created and opened an analyzer project, say CSharpAnalyzers in Visual Studio 2017. Refer to the first recipe in this chapter to create this project.

How to do it...

  1. In Solution Explorer, double-click on Resources.resx file in CSharpAnalyzers project to open the resource file in the resource editor.
  2. Replace the existing resource strings for AnalyzerDescription, AnalyzerMessageFormat and AnalyzerTitle with new strings:
  1. Replace the Initialize method implementation with the following:
public override void Initialize(AnalysisContext context)
{
context.RegisterSymbolAction(symbolContext =>
{
var symbolName = symbolContext.Symbol.Name;

// Skip the immediate containing type, CS0542 already covers this case.
var outerType = symbolContext.Symbol.ContainingType?.ContainingType;
while (outerType != null)
{
// Check if the current outer type has the same name as the given member.
if (symbolName.Equals(outerType.Name))
{
// For all such symbols, report a diagnostic.
var diagnostic = Diagnostic.Create(Rule, symbolContext.Symbol.Locations[0], symbolContext.Symbol.Name);
symbolContext.ReportDiagnostic(diagnostic);
return;
}

outerType = outerType.ContainingType;
}
},
SymbolKind.NamedType,
SymbolKind.Method,
SymbolKind.Field,
SymbolKind.Event,
SymbolKind.Property);
}
  1. Click on Ctrl + F5 to start a new Visual Studio instance with the analyzer enabled.
  2. In the new Visual Studio instance, create a new C# class library with the following code:
namespace ClassLibrary
{
public class OuterClass
{
public class NestedClass
{
public class NestedClass
{
}
}
}
}
  1. Verify the compiler reported diagnostic CS0542 in the error list: 'NestedClass': member names cannot be the same as their enclosing type.
  2. Change the class library code to following:
namespace ClassLibrary
{
public class OuterClass
{
public class NestedClass
{
public class InnerClass
{
public class NestedClass
{
}
}
}
}
}
  1. Verify that CS0542 isn t reported anymore, but the error list has our analyzer diagnostic:
  1. Replace the innermost type declaration for NestedClass with a field: public int NestedClass, and verify the same analyzer diagnostic is reported. You should get the same diagnostic for other member kinds such as method, property, and events with the same name.

How it works...

Symbol analyzers register one or more symbol action callbacks to analyze symbol kinds of interest. Note that, unlike the default implementation that registered a delegate method named AnalyzeSymbol, we registered a lambda callback.

We specified interest in analyzing all the top-level symbol kinds that can have an enclosing type, namely types, methods, fields, properties, and events in the RegisterSymbolAction invocation:

context.RegisterSymbolAction(symbolContext =>
{
...
},
SymbolKind.NamedType,
SymbolKind.Method,
SymbolKind.Field,
SymbolKind.Event,
SymbolKind.Property);

The analyzer driver ensures that the registered lambda is invoked for all symbols of the registered interest kinds in the compilation.

Analysis skips the immediate enclosing type, as C# compiler already reports error CS0542, if a member has the same name as its enclosing type.

// Skip the immediate containing type, CS0542 already covers this case.
var outerType = symbolContext.Symbol.ContainingType?.ContainingType;

Core analysis works by looping over the outer types and comparing the name of the symbol in a symbol analysis context with the relevant outer types, until it finds a match, in which case, it reports a diagnostic; if the outer type has no containing type, it doesn t report a diagnostic.

while (outerType != null)
{
// Check if the current outer type has the same name as the given member.
if (symbolName.Equals(outerType.Name))
{
// For all such symbols, report a diagnostic.
...
}

outerType = outerType.ContainingType;
}
It is recommended that symbol actions only analyze and report diagnostics about declarations, not the executable code within it. If you need to analyze executable code within a symbol, you should try to register other action kinds discussed later in this chapter.

There s more...

Trivia: The preceding implementation of the symbol analyzer does not have optimal performance. For example, if you have n levels of type nesting, and m fields in the inner innermost nested type, the analysis we implemented will be O(m*n) algorithmic complexity. Can you implement an alternate implementation where the analysis can be implemented with a much superior O(m + n) complexity?

See also

Our current analyzer implementation is completely stateless because it doesn t require analysis that is dependent upon more than one symbol at a time. We analyze each symbol individually and report diagnostics for it. However, if you need to do more complex analysis that requires collecting state from multiple symbols and then doing a compilation-wide analysis, you should write a stateful compilation analyzer with symbol and compilation actions. This is covered later in this chapter in the recipe Creating a compilation analyzer to analyze a whole compilation and report issues.

You have been reading a chapter from
Roslyn Cookbook
Published in: Jul 2017
Publisher: Packt
ISBN-13: 9781787286832
Register for a free Packt account to unlock a world of extra content!
A free Packt account unlocks extra newsletters, articles, discounted offers, and much more. Start advancing your knowledge today.
Unlock this book and the full library FREE for 7 days
Get unlimited access to 7000+ expert-authored eBooks and videos courses covering every tech area you can think of
Renews at ₹800/month. Cancel anytime