fix, UB in symlist, GC freed some strings in the unmanaged symbol tree

cesyms/cesyms.d

@@ -12,7 +12,7 @@ Usage
   - `<filename>`: optional, the D module filename, if not set then the program
     reads the module from stdin.
   - see the source for more information about how to use the output.
-    It's basically a tree of struct with 3 members: symbol type, name and location.
+    It's basically a tree of struct with 3 members: symbol type, name and location
 
 - Test in CE as a runnable module:
   click `Compile file and run ...` and type either `<CFF>` or `-j <CFF>` in the
@@ -37,115 +37,138 @@ void main(string[] args)
 {
     // format
     bool asJson;
-    getopt(args, std.getopt.config.passThrough,'j', &asJson); 
+    getopt(args, std.getopt.config.passThrough, 'j', &asJson);
 
     // get either the module from stdin or from first arg
     string fname;
     ubyte[] source;
     if (args.length == 1)
     {
-        version(runnable_module)
+        version (runnable_module)
         {
             source = cast(ubyte[]) read(__FILE__, size_t.max);
         }
-        else foreach(buff; stdin.byChunk(1024))
+        else
+            foreach (buff; stdin.byChunk(1024))
                 source ~= buff;
     }
     else if (args.length == 2)
     {
-        fname = args[$-1];
+        fname = args[$ - 1];
-        if (!fname.exists) return;
+        if (!fname.exists)
+            return;
         source = cast(ubyte[]) read(fname, size_t.max);
     }
-    else return;
+    else
+        return;
 
     // load and parse the file
     auto config = LexerConfig(fname, StringBehavior.source, WhitespaceBehavior.skip);
     auto scache = StringCache(StringCache.defaultBucketCount);
-    auto ast = parseModule(getTokensForParser(source, config, &scache), fname, null, &(SymbolListBuilder.astError));
+    auto ast = parseModule(getTokensForParser(source, config, &scache), fname,
+            null, &(SymbolListBuilder.astError));
 
     // visit each root member
     SymbolListBuilder slb = construct!SymbolListBuilder;
-    foreach(Declaration decl; ast.declarations)
+    foreach (Declaration decl; ast.declarations)
     {
         slb.resetRoot;
         slb.visit(decl);
     }
 
-    version(none)
+    version (none)
     {
         int level = -1;
-        void print(Symbol * s)
+        void print(Symbol* s)
         {
-            foreach(i; 0 .. level) write(".");
+            foreach (i; 0 .. level)
+                write(".");
             level++;
             write(s.name, '\r');
-            foreach(ss; s.subs)
+            foreach (ss; s.subs)
                 print(ss);
 
             level--;
         }
+
         print(slb.root);
     }
     else
     {
-        if (asJson) write(slb.serializeJson);
+        if (asJson)
-        else write(slb.serializePascal);
+            write(slb.serializeJson);
+        else
+            write(slb.serializePascal);
     }
 
     slb.destruct;
 }
 
 // libdparse warnings includes some "'", which in Pascal are string delim
-string patchPasStringLitteral(const ref string p)
+string patchPasStringLitteral(string p)
 {
+    import std.range : empty, front, popFront;
+
     string result;
-    for (auto i = 0; i < p.length; i++)
+    while (!p.empty)
     {
-        auto curr = p[i];
+        dchar curr = p.front;
-        if (curr == 0)
+        switch (curr)
-            break;
+        {
-        else if (curr == 13 || curr == 10)
+        default:
-            result ~= ' ';
-        else if (curr == '\'') 
-            result ~= "'#39'";
-        else  
             result ~= curr;
+            break;
+        case 10, 13:
+            result ~= ' ';
+            break;
+        case '\'':
+            result ~= "'#39'";
+        }
+        p.popFront;
     }
     return result;
 }
 
 // Memory utils ---------------------------------------------------------------+
-void * getMem(size_t size) nothrow
+void* getMem(size_t size) nothrow
 {
     import std.c.stdlib;
+
     auto result = malloc(size);
     assert(result, "Out of memory");
     return result;
 }
 
-CT construct(CT, A...)(A a) 
+CT construct(CT, A...)(A a) if (is(CT == class) && !isAbstractClass!CT)
-if (is(CT == class))
 {
-    import std.conv : emplace;
+    auto size = typeid(CT).init.length;
-    auto size = __traits(classInstanceSize, CT);
+    auto memory = getMem(size);
-    auto memory = getMem(size)[0 .. size];
+    memory[0 .. size] = typeid(CT).init[];
-    return emplace!(CT, A)(memory, a);
+    static if (__traits(hasMember, CT, "__ctor"))
+        (cast(CT)(memory)).__ctor(a);
+    import core.memory : GC;
+
+    GC.addRange(memory, size, typeid(CT));
+    return cast(CT) memory;
 }
 
-ST * construct(ST, A...)(A a)
+ST* construct(ST, A...)(A a) if (is(ST == struct) || is(ST == union))
-if(is(ST==struct))
 {
     import std.conv : emplace;
+
     auto size = ST.sizeof;
     auto memory = getMem(size)[0 .. size];
+    import core.memory : GC;
+
+    GC.addRange(memory.ptr, size, typeid(ST));
     return emplace!(ST, A)(memory, a);
 }
 
 void destruct(T)(ref T instance)
-if (is(T == class) || (isPointer!T && is(PointerTarget!T == struct)))
+        if (is(T == class) || (isPointer!T && is(PointerTarget!T == struct)))
 {
-    if (!instance) return;
+    if (!instance)
+        return;
     destroy(instance);
     instance = null;
 }
@@ -175,11 +198,11 @@ struct Symbol
     size_t col;
     string name;
     SymbolType type;
-    Symbol * [] subs;
+    Symbol*[] subs;
 
     ~this()
     {
-        foreach_reverse(i; 0..subs.length)
+        foreach_reverse (i; 0 .. subs.length)
             subs[i].destruct;
     }
 
@@ -189,7 +212,8 @@ struct Symbol
             serializePascal(list);
         else static if (is(List == JSONValue))
             serializeJson(list);
-        else static assert(0, "serialization kind cannot be deduced from list");
+        else
+            static assert(0, "serialization kind cannot be deduced from list");
     }
 
     void serializePascal(ref Appender!string lfmApp)
@@ -202,7 +226,8 @@ struct Symbol
         lfmApp.put(format("symType = %s\r", type));
 
         lfmApp.put("subs = <");
-        if (subs.length) foreach(Symbol * sub; subs)
+        if (subs.length)
+            foreach (Symbol* sub; subs)
                 sub.serialize(lfmApp);
         lfmApp.put(">\r");
         lfmApp.put("end");
@@ -211,14 +236,14 @@ struct Symbol
     void serializeJson(ref JSONValue json)
     {
         auto vobj = parseJSON("{}");
-        vobj["line"]= JSONValue(line);
+        vobj["line"] = JSONValue(line);
         vobj["col"] = JSONValue(col);
-        vobj["name"]= JSONValue(name);
+        vobj["name"] = JSONValue(name);
-        vobj["type"]= JSONValue(to!string(type));
+        vobj["type"] = JSONValue(to!string(type));
         if (subs.length)
         {
             auto vsubs = parseJSON("[]");
-            foreach(Symbol * sub; subs)
+            foreach (Symbol* sub; subs)
                 sub.serializeJson(vsubs);
             vobj["items"] = vsubs;
         }
@@ -230,8 +255,8 @@ struct Symbol
 // AST visitor/Symbol list ----------------------------------------------------+
 class SymbolListBuilder : ASTVisitor
 {
-    Symbol * root;
+    Symbol* root;
-    Symbol * parent;
+    Symbol* parent;
 
     // for some reason (?) the .name of a (static Symbol* []) item was lost
     __gshared static Symbol[] illFormed;
@@ -253,15 +278,18 @@ class SymbolListBuilder : ASTVisitor
 
     static void astError(string fname, size_t line, size_t col, string msg, bool isErr)
     {
-        Symbol * newSym = construct!Symbol;
+        Symbol* newSym = construct!Symbol;
         newSym.col = col;
         newSym.line = line;
         newSym.name = patchPasStringLitteral(msg);
         newSym.type = isErr ? SymbolType._error : SymbolType._warning;
-        illFormed ~= * newSym;
+        illFormed ~= *newSym;
     }
 
-    final void resetRoot(){parent = root;}
+    final void resetRoot()
+    {
+        parent = root;
+    }
 
     final string serializePascal()
     {
@@ -269,8 +297,10 @@ class SymbolListBuilder : ASTVisitor
         lfmApp.reserve(count * 64);
 
         lfmApp.put("object TSymbolList\rsymbols = <");
-        foreach(sym; illFormed) sym.serialize(lfmApp);
+        foreach (sym; illFormed)
-        foreach(sym; root.subs) sym.serialize(lfmApp);
+            sym.serialize(lfmApp);
+        foreach (sym; root.subs)
+            sym.serialize(lfmApp);
         lfmApp.put(">\rend\r\n");
 
         return lfmApp.data;
@@ -280,40 +310,34 @@ class SymbolListBuilder : ASTVisitor
     {
         JSONValue result = parseJSON("{}");
         JSONValue vsubs = parseJSON("[]");
-        foreach(sym; illFormed) sym.serialize(vsubs);
+        foreach (sym; illFormed)
-        foreach(sym; root.subs) sym.serialize(vsubs);
+            sym.serialize(vsubs);
+        foreach (sym; root.subs)
+            sym.serialize(vsubs);
         result["items"] = vsubs;
-        version(assert) return result.toPrettyString;
+        version (assert)
+            return result.toPrettyString;
         // else: release mode
-        else return result.toString;
+        else
+            return result.toString;
     }
 
-    /// returns a new symbol if the declarator is based on a Token named "name".
+    /// visitor implementation if the declaration has a "name".
-    final Symbol * addDeclaration(DT)(DT adt)
-    {
-        static if (__traits(hasMember, DT, "name"))
-        {
-            count++;
-            auto result = construct!Symbol;
-            result.name = adt.name.text;
-            result.line = adt.name.line;
-            result.col  = adt.name.column;
-            parent.subs ~= result;
-            return result;
-        }
-        else static assert(0, "addDeclaration no implemented for " ~ DT.stringof);
-    }
-    
-    /// visitor implementation if the declarator is based on a Token named "name".
     final void namedVisitorImpl(DT, SymbolType st, bool dig = true)(const(DT) dt)
+    if (__traits(hasMember, DT, "name"))
     {
-        auto newSymbol = addDeclaration(dt);
+        ++count;
+        Symbol* newSymbol = construct!Symbol;
+        newSymbol.name = dt.name.text;
+        newSymbol.line = dt.name.line;
+        newSymbol.col = dt.name.column;
         newSymbol.type = st;
-        //
+        parent.subs ~= newSymbol;
         static if (dig)
         {
             auto previousParent = parent;
-            scope(exit) parent = previousParent;
+            scope (exit)
+                parent = previousParent;
             parent = newSymbol;
             dt.accept(this);
         }
@@ -323,8 +347,8 @@ class SymbolListBuilder : ASTVisitor
     final void otherVisitorImpl(SymbolType st, string name, size_t line, size_t col)
     {
         count++;
-        auto result = construct!Symbol;
+        Symbol* result = construct!Symbol;
-        result.name = name;
+        result.name = name.idup;
         result.line = line;
         result.col = col;
         result.type = st;
@@ -333,8 +357,7 @@ class SymbolListBuilder : ASTVisitor
 
     final override void visit(const AliasDeclaration decl)
     {
-        // why is initializers an array ?
+        if (decl.initializers.length)
-        if (decl.initializers.length > 0)
             namedVisitorImpl!(AliasInitializer, SymbolType._alias)(decl.initializers[0]);
     }
 
@@ -350,9 +373,11 @@ class SymbolListBuilder : ASTVisitor
 
     final override void visit(const AutoDeclaration decl)
     {
-        otherVisitorImpl(SymbolType._enum, decl.identifiers[0].text,
+        if (decl.identifiers.length)
+        {
+            otherVisitorImpl(SymbolType._variable, decl.identifiers[0].text,
                 decl.identifiers[0].line, decl.identifiers[0].column);
-        decl.accept(this);
+        }
     }
 
     final override void visit(const ClassDeclaration decl)
@@ -392,22 +417,20 @@ class SymbolListBuilder : ASTVisitor
 
     final override void visit(const ImportDeclaration decl)
     {
-        foreach(const(SingleImport) si; decl.singleImports)
+        foreach (const(SingleImport) si; decl.singleImports)
         {
             if (!si.identifierChain.identifiers.length)
                 continue;
-            //
+
             string[] modules;
-            foreach(ident; si.identifierChain.identifiers)
+            foreach (ident; si.identifierChain.identifiers)
             {
                 modules ~= ident.text;
                 modules ~= ".";
             }
-            //
+            otherVisitorImpl(SymbolType._import, modules[0 .. $ - 1].join,
-            otherVisitorImpl(SymbolType._import, modules[0..$-1].join, 
                 si.identifierChain.identifiers[0].line,
-                si.identifierChain.identifiers[0].column
+                si.identifierChain.identifiers[0].column);
-            ); 
         }
     }
 
@@ -433,9 +456,10 @@ class SymbolListBuilder : ASTVisitor
 
     final override void visit(const VariableDeclaration decl)
     {
-        foreach(elem; decl.declarators)
+        if (decl.declarators)
+            foreach (elem; decl.declarators)
                 namedVisitorImpl!(Declarator, SymbolType._variable, false)(elem);
-        if (decl.autoDeclaration)
+        else if (decl.autoDeclaration)
             visit(decl.autoDeclaration);
     }
 
@@ -450,3 +474,4 @@ class SymbolListBuilder : ASTVisitor
     }
 }
 //----
+