(stdlib) Add basic StringBuilder class

...and use it everywhere instead of the .NET class. This can be used to construct Perlang strings with dynamic content. The `StringBuilder` class uses a byte buffer behind the scenes, allowing the string to be mutated until the user is "finished" with it. Once that happens, an (immutable) string is produced which is then carried along in the program. https://gitlab.perlang.org/perlang/perlang/-/merge_requests/548
perlang-org · Oct 30, 2024 · fa14166 · fa14166
1 parent 21ab6e7
commit fa14166
Show file tree

Hide file tree

Showing 19 changed files with 355 additions and 56 deletions.
diff --git a/.editorconfig b/.editorconfig
@@ -0,0 +1,5 @@
+root = true
+
+[*.cs]
+# CA2000: Dispose objects before losing scope
+dotnet_diagnostic.CA2000.severity = error
diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
@@ -54,6 +54,7 @@ test:
   script:
     - make auto-generated
     - make stdlib
+    - make perlang_cli_install_integration_test_release
     - >
         if [[ "$TEST_MODE" == 'valgrind' ]]; then
           PERLANG_RUN_WITH_VALGRIND=true dotnet test --configuration Release --verbosity minimal --logger "junit;LogFilePath=perlang-junit-log.xml";

diff --git a/Makefile b/Makefile
@@ -1,9 +1,9 @@
 # src/Perlang.Common/CommonConstants.Generated.cs is not phony, but we always want
 # to force it to be regenerated.
 .PHONY: \
-	all auto-generated clean darkerfx-push docs docs-serve docs-test-examples \
-	docs-validate-api-docs install install-latest-snapshot perlang_cli \
-	perlang_cli_clean perlang_cli_install_debug perlang_cli_install_release perlang_cli_install_integration_test_debug \
+	all auto-generated clean darkerfx-push docs docs-serve docs-test-examples valgrind-docs-test-examples \
+	docs-validate-api-docs install install-latest-snapshot perlang_cli perlang_cli_clean perlang_cli_install_debug \
+	perlang_cli_install_release perlang_cli_install_integration_test_debug perlang_cli_install_integration_test_release \
 	publish-release release run run-hello-world-example valgrind-perlang-run-hello-world-example valgrind-hello-world-example \
 	test test-stdlib src/Perlang.Common/CommonConstants.Generated.cs
 
@@ -61,6 +61,10 @@ docs-test-examples: perlang_cli_install_release
 	for e in docs/examples/quickstart/*.per ; do echo -e \\n\\e[1m[$$e]\\e[0m ; $(RELEASE_PERLANG) $$e ; done
 	for e in docs/examples/the-language/*.per ; do echo -e \\n\\e[1m[$$e]\\e[0m ; $(RELEASE_PERLANG) $$e ; done
 
+valgrind-docs-test-examples: perlang_cli_install_release
+	for e in docs/examples/quickstart/*.per ; do echo -e \\n\\e[1m[$$e]\\e[0m ; $(VALGRIND) $(DEBUG_PERLANG) $$e ; done
+	for e in docs/examples/the-language/*.per ; do echo -e \\n\\e[1m[$$e]\\e[0m ; $(VALGRIND) $(DEBUG_PERLANG) $$e ; done
+
 docs-serve:
 	live-server _site
 
@@ -142,6 +146,12 @@ perlang_cli_install_integration_test_debug: perlang_cli
 	mkdir -p src/Perlang.Tests.Integration/bin/Debug/net8.0
 	cp lib/perlang_cli/lib/perlang_cli.so src/Perlang.Tests.Integration/bin/Debug/net8.0
 
+perlang_cli_install_integration_test_release: perlang_cli
+	mkdir -p src/Perlang.Tests/bin/Release/net8.0
+	mkdir -p src/Perlang.Tests.Integration/bin/Release/net8.0
+	cp lib/perlang_cli/lib/perlang_cli.so src/Perlang.Tests/bin/Release/net8.0
+	cp lib/perlang_cli/lib/perlang_cli.so src/Perlang.Tests.Integration/bin/Release/net8.0
+
 test:
 	dotnet test --configuration Release
 

diff --git a/global.ruleset b/global.ruleset
@@ -46,6 +46,7 @@
         <Rule Id="S3260" Action="None"/>        <!-- Private classes or records which are not derived in the current assembly should be marked as 'sealed' -->
         <Rule Id="S3267" Action="None"/>        <!-- Loops should be simplified with "LINQ" expressions -->
         <Rule Id="S3376" Action="None"/>
+        <Rule Id="S3881" Action="None"/>        <!-- "IDisposable" should be implemented correctly -->
         <Rule Id="S3925" Action="None"/>
     </Rules>
 </RuleSet>
diff --git a/src/Perlang.ConsoleApp/Program.cs b/src/Perlang.ConsoleApp/Program.cs
@@ -8,7 +8,6 @@
 using System.CommandLine.Parsing;
 using System.IO;
 using System.Linq;
-using System.Text;
 using System.Threading.Tasks;
 using Perlang.Compiler;
 using Perlang.Interpreter;
@@ -24,7 +23,7 @@
 
 namespace Perlang.ConsoleApp
 {
-    public class Program
+    public class Program : IDisposable
     {
         internal enum ExitCodes
         {
@@ -179,7 +178,7 @@ public static int Main(string[] args)
                         string? outputFileName = parseResult.GetValueForOption(outputOption);
                         bool idempotent = parseResult.GetValueForOption(idempotentOption);
 
-                        var program = new Program(
+                        using var program = new Program(
                             replMode: false,
                             standardOutputHandler: Console.WriteLine,
                             disabledWarningsAsErrors: disabledWarningsAsErrorsList,
@@ -197,7 +196,7 @@ public static int Main(string[] args)
 
                         if (parseResult.Tokens.Count == 1)
                         {
-                            var program = new Program(
+                            using var program = new Program(
                                 replMode: false,
                                 standardOutputHandler: Console.WriteLine,
                                 disabledWarningsAsErrors: disabledWarningsAsErrorsList,
@@ -210,7 +209,7 @@ public static int Main(string[] args)
                         {
                             var remainingArguments = parseResult.GetValueForArgument(scriptArguments);
 
-                            var program = new Program(
+                            using var program = new Program(
                                 replMode: false,
                                 arguments: remainingArguments,
                                 standardOutputHandler: Console.WriteLine,
@@ -241,7 +240,7 @@ public static int Main(string[] args)
                 .Invoke(args);
         }
 
-        internal Program(
+        private Program(
             bool replMode,
             Action<Lang.String> standardOutputHandler,
             IEnumerable<string>? arguments = null,
@@ -265,6 +264,11 @@ internal Program(
             );
         }
 
+        public void Dispose()
+        {
+            compiler.Dispose();
+        }
+
         private int RunFile(string path, string? outputPath)
         {
             if (!File.Exists(path))
@@ -293,7 +297,7 @@ private int RunFile(string path, string? outputPath)
 
         private int CompileAndAssembleFile(string[] scriptFiles, string? targetPath, bool idempotent)
         {
-            var completeSource = new StringBuilder();
+            using var completeSource = NativeStringBuilder.Create();
 
             foreach (string scriptFile in scriptFiles)
             {

diff --git a/src/Perlang.Interpreter/Compiler/PerlangCompiler.cs b/src/Perlang.Interpreter/Compiler/PerlangCompiler.cs
@@ -11,7 +11,6 @@
 using System.Linq;
 using System.Numerics;
 using System.Reflection;
-using System.Text;
 using Perlang.Attributes;
 using Perlang.Compiler;
 using Perlang.Exceptions;
@@ -22,6 +21,7 @@
 using Perlang.Interpreter.NameResolution;
 using Perlang.Interpreter.Typing;
 using Perlang.Lang;
+using Perlang.Native;
 using Perlang.Parser;
 using Perlang.Stdlib;
 using static Perlang.TokenType;
@@ -58,7 +58,7 @@ namespace Perlang.Interpreter.Compiler;
 /// Also, this class is not thread safe. A single instance of this class cannot safely be used from multiple
 /// threads.</remarks>
 /// </summary>
-public class PerlangCompiler : Expr.IVisitor<object?>, Stmt.IVisitor<object>
+public class PerlangCompiler : Expr.IVisitor<object?>, Stmt.IVisitor<object>, IDisposable
 {
     /// <summary>
     /// Gets a value indicating whether caching of compiled results should be disabled or not.
@@ -106,7 +106,7 @@ private static bool CompilationCacheDisabled
     /// <summary>
     /// The location in the output we are currently writing to. Can be an enum, function, etc.
     /// </summary>
-    private readonly StringBuilder mainMethodContent;
+    private readonly NativeStringBuilder mainMethodContent;
 
     private int indentationLevel = 1;
 
@@ -128,7 +128,8 @@ public PerlangCompiler(
         this.standardOutputHandler = standardOutputHandler;
         this.BindingHandler = bindingHandler ?? new BindingHandler();
 
-        this.mainMethodContent = new StringBuilder();
+        this.mainMethodContent = NativeStringBuilder.Create();
+
         this.methods = new Dictionary<string, Method>();
         this.enums = new Dictionary<string, string>();
 
@@ -142,6 +143,15 @@ public PerlangCompiler(
         nativeClasses = RegisterGlobalFunctionsAndClasses();
     }
 
+    public void Dispose()
+    {
+        mainMethodContent.Dispose();
+
+        foreach (Method method in methods.Values) {
+            method.MethodBody.Dispose();
+        }
+    }
+
     private IImmutableDictionary<string, Type> CreateSuperGlobals(ImmutableList<string> argumentsList)
     {
         // Set up the super-global ARGV variable.
@@ -824,7 +834,7 @@ private void RegisterGlobalClasses()
     /// <returns>The C++ source for the given statements.</returns>
     private string Compile(IEnumerable<Stmt> statements)
     {
-        var result = new StringBuilder();
+        using var result = NativeStringBuilder.Create();
 
         foreach (Stmt statement in statements)
         {
@@ -857,7 +867,7 @@ public object VisitAssignExpr(Expr.Assign expr)
 
     public object VisitBinaryExpr(Expr.Binary expr)
     {
-        var result = new StringBuilder();
+        using var result = NativeStringBuilder.Create();
         string? leftCast = null;
         string? rightCast = null;
 
@@ -1206,7 +1216,7 @@ public object VisitCallExpr(Expr.Call expr)
         // TODO: binding at an earlier stage and e.g. validate the number and type of arguments with the method
         // TODO: definition. Right now, we'll leave this to the C++ compiler to deal with...
 
-        var result = new StringBuilder();
+        using var result = NativeStringBuilder.Create();
 
         result.Append(expr.Callee.Accept(this));
         result.Append('(');
@@ -1237,7 +1247,7 @@ public object VisitIndexExpr(Expr.Index expr)
         // TODO: One way to deal with this: make `foo[bar]` be safe and add an "unsafe" indexing operator `foo[[bar]]`
         // TODO: which can be used when the user believes he knows best. Investigate how Rust is doing this.
 
-        var result = new StringBuilder();
+        using var result = NativeStringBuilder.Create();
 
         ITypeReference indexeeTypeReference = expr.Indexee.TypeReference;
 
@@ -1262,7 +1272,7 @@ public object VisitIndexExpr(Expr.Index expr)
 
     public object VisitGroupingExpr(Expr.Grouping expr)
     {
-        var result = new StringBuilder();
+        using var result = NativeStringBuilder.Create();
 
         result.Append('(');
         result.Append(expr.Expression.Accept(this));
@@ -1273,7 +1283,7 @@ public object VisitGroupingExpr(Expr.Grouping expr)
 
     public object VisitCollectionInitializerExpr(Expr.CollectionInitializer collectionInitializer)
     {
-        var result = new StringBuilder();
+        using var result = NativeStringBuilder.Create();
 
         result.Append($"std::make_shared<{collectionInitializer.TypeReference.CppType}>(");
 
@@ -1309,7 +1319,7 @@ public object VisitCollectionInitializerExpr(Expr.CollectionInitializer collecti
 
     public object VisitLiteralExpr(Expr.Literal expr)
     {
-        var result = new StringBuilder();
+        using var result = NativeStringBuilder.Create();
 
         if (expr.Value is AsciiString)
         {
@@ -1393,7 +1403,7 @@ public object VisitLiteralExpr(Expr.Literal expr)
 
     public object VisitLogicalExpr(Expr.Logical expr)
     {
-        var result = new StringBuilder();
+        using var result = NativeStringBuilder.Create();
 
         switch (expr.Operator.Type)
         {
@@ -1419,7 +1429,7 @@ public object VisitLogicalExpr(Expr.Logical expr)
 
     public object VisitUnaryPrefixExpr(Expr.UnaryPrefix expr)
     {
-        var result = new StringBuilder();
+        using var result = NativeStringBuilder.Create();
 
         switch (expr.Operator.Type)
         {
@@ -1443,7 +1453,7 @@ public object VisitUnaryPrefixExpr(Expr.UnaryPrefix expr)
 
     public object VisitUnaryPostfixExpr(Expr.UnaryPostfix expr)
     {
-        var result = new StringBuilder();
+        using var result = NativeStringBuilder.Create();
 
         switch (expr.Operator.Type)
         {
@@ -1470,7 +1480,7 @@ public object VisitIdentifierExpr(Expr.Identifier expr) =>
 
     public object VisitGetExpr(Expr.Get expr)
     {
-        var result = new StringBuilder();
+        using var result = NativeStringBuilder.Create();
 
         if (expr.Object is Expr.Identifier identifier)
         {
@@ -1530,7 +1540,7 @@ public object VisitGetExpr(Expr.Get expr)
 
     public object VisitBlockStmt(Stmt.Block block)
     {
-        var result = new StringBuilder();
+        using var result = NativeStringBuilder.Create();
 
         result.Append(Indent(indentationLevel));
         result.AppendLine("{");
@@ -1555,7 +1565,8 @@ public object VisitClassStmt(Stmt.Class stmt)
 
     public object VisitEnumStmt(Stmt.Enum stmt)
     {
-        var stringBuilder = new StringBuilder();
+        using var stringBuilder = NativeStringBuilder.Create();
+
         int localIndentationLevel = 0;
 
         // The "enum class" concept as defined in C++11 would be preferable here since it provides better type safety, but
@@ -1571,8 +1582,7 @@ public object VisitEnumStmt(Stmt.Enum stmt)
         stringBuilder.AppendLine($"enum {stmt.Name.Lexeme} {{");
         localIndentationLevel++;
 
-        foreach (KeyValuePair<string, Expr?> enumMember in stmt.Members)
-        {
+        foreach (KeyValuePair<string, Expr?> enumMember in stmt.Members) {
             stringBuilder.Append(Indent(localIndentationLevel));
             stringBuilder.Append(enumMember.Key);
 
@@ -1594,15 +1604,15 @@ public object VisitEnumStmt(Stmt.Enum stmt)
 
         enums[stmt.Name.Lexeme] = stringBuilder.ToString();
 
-        // Does not need to return the StringBuilder here, since it's been stored in the methods dictionary already.
+        // Does not need to return the StringBuilder here, since it's been stored in the enums dictionary already.
         return VoidObject.Void;
     }
 
     public object VisitExpressionStmt(Stmt.ExpressionStmt stmt)
     {
-        var result = new StringBuilder();
+        using var result = NativeStringBuilder.Create();
 
-        // We only emit the "wrapping" (whitespace before + semicolon and newline after expression)  in this case; the
+        // We only emit the "wrapping" (whitespace before + semicolon and newline after expression) in this case; the
         // rest comes from visiting the expression itself
         result.Append(Indent(indentationLevel));
         result.Append(stmt.Expression.Accept(this));
@@ -1613,7 +1623,10 @@ public object VisitExpressionStmt(Stmt.ExpressionStmt stmt)
 
     public object VisitFunctionStmt(Stmt.Function functionStmt)
     {
-        var functionContent = new StringBuilder();
+        // Disposed via Dispose() method for class
+#pragma warning disable CA2000
+        var functionContent = NativeStringBuilder.Create();
+#pragma warning restore CA2000
 
         if (methods.ContainsKey(functionStmt.Name.Lexeme))
         {
@@ -1641,7 +1654,7 @@ public object VisitFunctionStmt(Stmt.Function functionStmt)
 
     public object VisitIfStmt(Stmt.If stmt)
     {
-        var result = new StringBuilder();
+        using var result = NativeStringBuilder.Create();
 
         result.Append(Indent(indentationLevel));
         result.Append($"if ({stmt.Condition.Accept(this)}) ");
@@ -1678,7 +1691,7 @@ public object VisitPrintStmt(Stmt.Print stmt) =>
 
     public object VisitReturnStmt(Stmt.Return stmt)
     {
-        var result = new StringBuilder();
+        using var result = NativeStringBuilder.Create();
 
         if (stmt.Value == null)
         {
@@ -1697,7 +1710,7 @@ public object VisitReturnStmt(Stmt.Return stmt)
 
     public object VisitVarStmt(Stmt.Var stmt)
     {
-        var result = new StringBuilder();
+        using var result = NativeStringBuilder.Create();
 
         string variableName = stmt.Name.Lexeme;
 
@@ -1717,7 +1730,7 @@ public object VisitVarStmt(Stmt.Var stmt)
 
     public object VisitWhileStmt(Stmt.While whileStmt)
     {
-        var result = new StringBuilder();
+        using var result = NativeStringBuilder.Create();
 
         result.Append(Indent(indentationLevel));
         result.Append($"while ({whileStmt.Condition.Accept(this)}) ");
@@ -1729,7 +1742,7 @@ public object VisitWhileStmt(Stmt.While whileStmt)
 
     private static string Indent(int level) => String.Empty.PadLeft(level * 4);
 
-    private record Method(string Name, IImmutableList<Parameter> Parameters, string ReturnType, StringBuilder MethodBody)
+    private record Method(string Name, IImmutableList<Parameter> Parameters, string ReturnType, NativeStringBuilder MethodBody)
     {
         /// <summary>
         /// Gets the method parameters as a comma-separated string, in C++ format. Example: `int foo, int bar`.