Add string debugging

compiler/Generator.cs

@@ -102,6 +102,7 @@ public class Generator
                          #include <stdint.h>
                          #include <stdlib.h>
                          #include <string.h>
+                         #include <stdio.h>
 
                          """);
 
@@ -148,6 +149,13 @@ public class Generator
                 (
                     """
                     #define FLAG_STRING_LITERAL 1
+                    #define STRING_DEBUG 1
+
+                    #if STRING_DEBUG
+                    #define STR_DBG(fmt, ...) fprintf(stderr, "[STR] " fmt "\n", ##__VA_ARGS__)
+                    #else
+                    #define STR_DBG(fmt, ...)
+                    #endif
 
                     struct nub_core_string
                     {
@@ -157,54 +165,114 @@ public class Generator
                         uint32_t flags;
                     };
 
+                    #if STRING_DEBUG
+                    static size_t nub_core_string_allocs = 0;
+                    static size_t nub_core_string_frees = 0;
+
+                    __attribute__((destructor))
+                    static void string_debug_report(void)
+                    {
+                        fprintf(stderr, "[STR] REPORT allocs=%zu frees=%zu leaks=%zu\n", nub_core_string_allocs, nub_core_string_frees, nub_core_string_allocs - nub_core_string_frees);
+                    }
+                    #endif
+
                     static inline void string_rc_inc(struct nub_core_string *string)
                     {
+                        if (string == NULL)
+                        {
+                            STR_DBG("INC null string");
+                            return;
+                        }
+
                         if (string->flags & FLAG_STRING_LITERAL)
                             return;
 
                         string->ref += 1;
+
+                        STR_DBG("INC: str=%p ref=%u \"%s\"", (void*)string, string->ref, string->data);
                     }
 
                     static inline void string_rc_dec(struct nub_core_string *string)
                     {
                         if (string == NULL)
+                        {
+                            STR_DBG("DEC null string");
                             return;
+                        }
 
                         if (string->flags & FLAG_STRING_LITERAL)
                             return;
 
+                        if (string->ref == 0)
+                        {
+                            STR_DBG("ERROR: DEC on zero refcount str=%p", (void*)string);
+                            return;
+                        }
+
                         string->ref -= 1;
 
+                        STR_DBG("DEC: str=%p ref=%u \"%s\"", (void*)string, string->ref, string->data);
+
                         if (string->ref == 0)
                         {
-                            free(string->data);
+                            STR_DBG("FREE: str=%p data=%p \"%s\"", (void*)string, (void*)string->data, string->data);
-                            free(string);
+
+                            char *data_ptr = string->data;
+                            struct nub_core_string *str_ptr = string;
+
+                        #if STRING_DEBUG
+                            nub_core_string_frees++;
+
+                            memset(data_ptr, 0xDD, str_ptr->length);
+                            memset(str_ptr, 0xDD, sizeof(*str_ptr));
+                        #endif
+
+                            free(data_ptr);
+                            free(str_ptr);
                         }
                     }
 
                     static inline struct nub_core_string *string_concat(struct nub_core_string *left, struct nub_core_string *right)
                     {
                         size_t new_length = left->length + right->length;
+
                         struct nub_core_string *result = malloc(sizeof(struct nub_core_string));
                         result->data = malloc(new_length + 1);
+
                         memcpy(result->data, left->data, left->length);
                         memcpy(result->data + left->length, right->data, right->length);
+
                         result->data[new_length] = '\0';
                         result->length = new_length;
                         result->ref = 1;
                         result->flags = 0;
+
+                    #if STRING_DEBUG
+                        nub_core_string_allocs++;
+                        STR_DBG("NEW concat: str=%p ref=%u \"%s\"", (void*)result, result->ref, result->data);
+                    #endif
+
                         return result;
                     }
 
-                    static inline struct nub_core_string *string_from_cstr(const char* cstr)
+                    static inline struct nub_core_string *string_from_cstr(char* cstr)
                     {
                         size_t len = strlen(cstr);
+
                         struct nub_core_string *result = malloc(sizeof(struct nub_core_string));
                         result->data = malloc(len + 1);
+
                         memcpy(result->data, cstr, len + 1);
+
                         result->length = len;
                         result->ref = 1;
                         result->flags = 0;
+
+                    #if STRING_DEBUG
+                        nub_core_string_allocs++;
+                        STR_DBG("NEW from_cstr: str=%p ref=%u \"%s\"", (void*)result, result->ref, result->data);
+                    #endif
+
                         return result;
                     }
 
@@ -491,6 +559,7 @@ public class Generator
             TypedNodeExpressionStringLiteral expression => EmitExpressionStringLiteral(expression),
             TypedNodeExpressionStructLiteral expression => EmitExpressionStructLiteral(expression),
             TypedNodeExpressionEnumLiteral expression => EmitExpressionEnumLiteral(expression),
+            TypedNodeExpressionStringConstructor expression => EmitExpressionStringConstructor(expression),
             TypedNodeExpressionStructMemberAccess expression => EmitExpressionMemberAccess(expression),
             TypedNodeExpressionStringLength expression => EmitExpressionStringLength(expression),
             TypedNodeExpressionStringPointer expression => EmitExpressionStringPointer(expression),
@@ -562,7 +631,7 @@ public class Generator
     {
         var name = Tmp();
         referencedStringLiterals.Add(name, expression.Value.Value);
-        return $"&{name}";
+        return $"(&{name})";
     }
 
     private string EmitExpressionStructLiteral(TypedNodeExpressionStructLiteral expression)
@@ -616,6 +685,15 @@ public class Generator
         return name;
     }
 
+    private string EmitExpressionStringConstructor(TypedNodeExpressionStringConstructor expression)
+    {
+        var name = Tmp();
+        scopes.Peek().DeconstructableNames.Add((name, expression.Type));
+        var value = EmitExpression(expression.Value);
+        writer.WriteLine($"{CType(expression.Type, name)} = string_from_cstr({value});");
+        return name;
+    }
+
     private string EmitExpressionMemberAccess(TypedNodeExpressionStructMemberAccess expression)
     {
         var target = EmitExpression(expression.Target);

compiler/Parser.cs

@@ -409,30 +409,45 @@ public class Parser
         {
             var type = ParseType();
 
-            if (TryExpectSymbol(Symbol.OpenParen))
+            if (type is NodeTypeString)
             {
+                ExpectSymbol(Symbol.OpenParen);
                 var value = ParseExpression();
                 ExpectSymbol(Symbol.CloseParen);
 
-                expr = new NodeExpressionEnumLiteral(TokensFrom(startIndex), type, value);
+                expr = new NodeExpressionStringConstructor(TokensFrom(startIndex), value);
             }
-            else if (TryExpectSymbol(Symbol.OpenCurly))
+            else if (type is NodeTypeNamed namedType)
             {
-                var initializers = new List<NodeExpressionStructLiteral.Initializer>();
+                if (TryExpectSymbol(Symbol.OpenParen))
-                while (!TryExpectSymbol(Symbol.CloseCurly))
                 {
-                    var initializerStartIndex = startIndex;
+                    var value = ParseExpression();
-                    var fieldName = ExpectIdent();
+                    ExpectSymbol(Symbol.CloseParen);
-                    ExpectSymbol(Symbol.Equal);
-                    var fieldValue = ParseExpression();
-                    initializers.Add(new NodeExpressionStructLiteral.Initializer(TokensFrom(initializerStartIndex), fieldName, fieldValue));
-                }
 
-                expr = new NodeExpressionStructLiteral(TokensFrom(startIndex), null, initializers);
+                    expr = new NodeExpressionEnumLiteral(TokensFrom(startIndex), namedType, value);
+                }
+                else if (TryExpectSymbol(Symbol.OpenCurly))
+                {
+                    var initializers = new List<NodeExpressionStructLiteral.Initializer>();
+                    while (!TryExpectSymbol(Symbol.CloseCurly))
+                    {
+                        var initializerStartIndex = startIndex;
+                        var fieldName = ExpectIdent();
+                        ExpectSymbol(Symbol.Equal);
+                        var fieldValue = ParseExpression();
+                        initializers.Add(new NodeExpressionStructLiteral.Initializer(TokensFrom(initializerStartIndex), fieldName, fieldValue));
+                    }
+
+                    expr = new NodeExpressionStructLiteral(TokensFrom(startIndex), null, initializers);
+                }
+                else
+                {
+                    expr = new NodeExpressionEnumLiteral(TokensFrom(startIndex), namedType, null);
+                }
             }
             else
             {
-                expr = new NodeExpressionEnumLiteral(TokensFrom(startIndex), type, null);
+                throw BasicError($"Expected named type or string", type);
             }
         }
         else
@@ -740,6 +755,11 @@ public class Parser
     {
         return new CompileException(Diagnostic.Error(message).At(fileName, ident).Build());
     }
+
+    private CompileException BasicError(string message, Node? node)
+    {
+        return new CompileException(Diagnostic.Error(message).At(fileName, node).Build());
+    }
 }
 
 public class Ast(string fileName, TokenIdent moduleName, List<NodeDefinition> definitions)
@@ -893,9 +913,9 @@ public class NodeExpressionBoolLiteral(List<Token> tokens, TokenBoolLiteral valu
     public TokenBoolLiteral Value { get; } = value;
 }
 
-public class NodeExpressionStructLiteral(List<Token> tokens, NodeType? type, List<NodeExpressionStructLiteral.Initializer> initializers) : NodeExpression(tokens)
+public class NodeExpressionStructLiteral(List<Token> tokens, NodeTypeNamed? type, List<NodeExpressionStructLiteral.Initializer> initializers) : NodeExpression(tokens)
 {
-    public NodeType? Type { get; } = type;
+    public NodeTypeNamed? Type { get; } = type;
     public List<Initializer> Initializers { get; } = initializers;
 
     public class Initializer(List<Token> tokens, TokenIdent name, NodeExpression value) : Node(tokens)
@@ -905,12 +925,17 @@ public class NodeExpressionStructLiteral(List<Token> tokens, NodeType? type, Lis
     }
 }
 
-public class NodeExpressionEnumLiteral(List<Token> tokens, NodeType type, NodeExpression? value) : NodeExpression(tokens)
+public class NodeExpressionEnumLiteral(List<Token> tokens, NodeTypeNamed type, NodeExpression? value) : NodeExpression(tokens)
 {
-    public NodeType Type { get; } = type;
+    public NodeTypeNamed Type { get; } = type;
     public NodeExpression? Value { get; } = value;
 }
 
+public class NodeExpressionStringConstructor(List<Token> tokens, NodeExpression value) : NodeExpression(tokens)
+{
+    public NodeExpression Value { get; } = value;
+}
+
 public class NodeExpressionMemberAccess(List<Token> tokens, NodeExpression target, TokenIdent name) : NodeExpression(tokens)
 {
     public NodeExpression Target { get; } = target;

compiler/TypeChecker.cs

@@ -263,6 +263,7 @@ public class TypeChecker
             NodeExpressionStringLiteral expression => CheckExpressionStringLiteral(expression, expectedType),
             NodeExpressionStructLiteral expression => CheckExpressionStructLiteral(expression, expectedType),
             NodeExpressionEnumLiteral expression => CheckExpressionEnumLiteral(expression, expectedType),
+            NodeExpressionStringConstructor expression => CheckExpressionStringConstructor(expression, expectedType),
             _ => throw new ArgumentOutOfRangeException(nameof(node))
         };
     }
@@ -651,6 +652,17 @@ public class TypeChecker
         return new TypedNodeExpressionEnumLiteral(expression.Tokens, type, value);
     }
 
+    private TypedNodeExpressionStringConstructor CheckExpressionStringConstructor(NodeExpressionStringConstructor expression, NubType? expectedType)
+    {
+        var stringPoitnerType = NubTypePointer.Get(NubTypeUInt.Get(8));
+
+        var value = CheckExpression(expression.Value, stringPoitnerType);
+        if (!value.Type.IsAssignableTo(stringPoitnerType))
+            throw BasicError($"Value of string constructor must be assignable to {stringPoitnerType}", value);
+
+        return new TypedNodeExpressionStringConstructor(expression.Tokens, NubTypeString.Instance, value);
+    }
+
     private NubType ResolveType(NodeType node)
     {
         return node switch
@@ -932,6 +944,11 @@ public class TypedNodeExpressionEnumLiteral(List<Token> tokens, NubType type, Ty
     public TypedNodeExpression? Value { get; } = value;
 }
 
+public class TypedNodeExpressionStringConstructor(List<Token> tokens, NubType type, TypedNodeExpression value) : TypedNodeExpression(tokens, type)
+{
+    public TypedNodeExpression Value { get; } = value;
+}
+
 public class TypedNodeExpressionStructMemberAccess(List<Token> tokens, NubType type, TypedNodeExpression target, TokenIdent name) : TypedNodeExpression(tokens, type)
 {
     public TypedNodeExpression Target { get; } = target;

examples/program/main.nub

@@ -16,6 +16,10 @@ func main(): i32 {
 
     let message = getMessage()
 
+    let newStr = new string("cstring".ptr)
+
+    core::println(newStr)
+
     match message {
         Quit {
             core::println("quit")