Merge pull request #412 from toshok/cache_module_str_constants

Conflicts:
	src/codegen/ast_interpreter.cpp
	src/runtime/types.cpp

src/codegen/ast_interpreter.cpp

@@ -645,7 +645,11 @@ Value ASTInterpreter::visit_stmt(AST_stmt* node) {
         case AST_TYPE::Exec:
             return visit_exec((AST_Exec*)node);
         case AST_TYPE::Expr:
+            // docstrings are str constant expression statements.
+            // ignore those while interpreting.
+            if ((((AST_Expr*)node)->value)->type != AST_TYPE::Str)
                 return visit_expr((AST_Expr*)node);
+            break;
         case AST_TYPE::Pass:
             return Value(); // nothing todo
         case AST_TYPE::Print:
@@ -1066,7 +1070,7 @@ Value ASTInterpreter::visit_set(AST_Set* node) {
 
 Value ASTInterpreter::visit_str(AST_Str* node) {
     if (node->str_type == AST_Str::STR) {
-        return boxString(node->str_data);
+        return source_info->parent_module->getStringConstant(node->str_data);
     } else if (node->str_type == AST_Str::UNICODE) {
         return decodeUTF8StringPtr(&node->str_data);
     } else {

src/codegen/irgen/irgenerator.cpp

@@ -1020,7 +1020,11 @@ private:
 
     CompilerVariable* evalStr(AST_Str* node, UnwindInfo unw_info) {
         if (node->str_type == AST_Str::STR) {
-            return makeStr(&node->str_data);
+            llvm::Value* rtn = embedConstantPtr(
+                irstate->getSourceInfo()->parent_module->getStringConstant(node->str_data), g.llvm_value_type_ptr);
+
+            return new ConcreteCompilerVariable(STR, rtn, true);
+
         } else if (node->str_type == AST_Str::UNICODE) {
             return makeUnicode(emitter, &node->str_data);
         } else {
@@ -2039,6 +2043,7 @@ private:
                 doExec(ast_cast<AST_Exec>(node), unw_info);
                 break;
             case AST_TYPE::Expr:
+                if ((((AST_Expr*)node)->value)->type != AST_TYPE::Str)
                     doExpr(ast_cast<AST_Expr>(node), unw_info);
                 break;
             // case AST_TYPE::If:

src/runtime/types.cpp

@@ -390,6 +390,25 @@ std::string BoxedModule::name() {
     }
 }
 
+Box* BoxedModule::getStringConstant(const std::string& ast_str) {
+    auto idx_iter = str_const_index.find(ast_str);
+    if (idx_iter != str_const_index.end())
+        return str_constants[idx_iter->second];
+
+    Box* box = boxString(ast_str);
+    str_const_index[ast_str] = str_constants.size();
+    str_constants.push_back(box);
+    return box;
+}
+
+extern "C" void moduleGCHandler(GCVisitor* v, Box* b) {
+    boxGCHandler(v, b);
+
+    BoxedModule* d = (BoxedModule*)b;
+
+    v->visitRange((void* const*)&d->str_constants[0], (void* const*)&d->str_constants[d->str_constants.size()]);
+}
+
 // This mustn't throw; our IR generator generates calls to it without "invoke" even when there are exception handlers /
 // finally-blocks in scope.
 // TODO: should we use C++11 `noexcept' here?
@@ -1890,8 +1909,8 @@ void setupRuntime() {
     function_cls->simple_destructor = builtin_function_or_method_cls->simple_destructor = functionDtor;
 
 
-    module_cls = new BoxedHeapClass(object_cls, NULL, offsetof(BoxedModule, attrs), 0, sizeof(BoxedModule), false,
-                                    new BoxedString("module"));
+    module_cls = new BoxedHeapClass(object_cls, &moduleGCHandler, offsetof(BoxedModule, attrs), 0, sizeof(BoxedModule),
+                                    false, new BoxedString("module"));
     member_cls
         = new BoxedHeapClass(object_cls, NULL, 0, 0, sizeof(BoxedMemberDescriptor), false, new BoxedString("member"));
     capifunc_cls

src/runtime/types.h

@@ -529,6 +529,11 @@ public:
     BoxedModule(const std::string& name, const std::string& fn, const char* doc = NULL);
     std::string name();
 
+    Box* getStringConstant(const std::string& ast_str);
+
+    llvm::StringMap<int> str_const_index;
+    std::vector<Box*> str_constants;
+
     DEFAULT_CLASS(module_cls);
 };