[clang] [clang-tools-extra] [clang-tidy] Add new check `modernize-use-designated-initializers` (PR #80541)
Danny Mösch via cfe-commits
cfe-commits at lists.llvm.org
Mon Feb 12 13:29:48 PST 2024
================
@@ -0,0 +1,127 @@
+//===--- UseDesignatedInitializersCheck.cpp - clang-tidy ------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "UseDesignatedInitializersCheck.h"
+#include "clang/AST/APValue.h"
+#include "clang/AST/Decl.h"
+#include "clang/AST/Expr.h"
+#include "clang/AST/Stmt.h"
+#include "clang/ASTMatchers/ASTMatchFinder.h"
+#include "clang/ASTMatchers/ASTMatchers.h"
+#include "clang/ASTMatchers/ASTMatchersMacros.h"
+#include "clang/Basic/Diagnostic.h"
+#include "clang/Lex/Lexer.h"
+#include "clang/Tooling/DesignatedInitializers.h"
+
+using namespace clang::ast_matchers;
+
+namespace clang::tidy::modernize {
+
+static const char *IgnoreSingleElementAggregatesName =
+ "IgnoreSingleElementAggregates";
+static const bool IgnoreSingleElementAggregatesDefault = true;
+
+static const char *RestrictToPODTypesName = "RestrictToPODTypes";
+static const bool RestrictToPODTypesDefault = false;
+
+UseDesignatedInitializersCheck::UseDesignatedInitializersCheck(
+ StringRef Name, ClangTidyContext *Context)
+ : ClangTidyCheck(Name, Context), IgnoreSingleElementAggregates(Options.get(
+ IgnoreSingleElementAggregatesName,
+ IgnoreSingleElementAggregatesDefault)),
+ RestrictToPODTypes(
+ Options.get(RestrictToPODTypesName, RestrictToPODTypesDefault)) {}
+
+AST_MATCHER(CXXRecordDecl, isAggregate) { return Node.isAggregate(); }
+
+AST_MATCHER(CXXRecordDecl, isPOD) { return Node.isPOD(); }
+
+AST_MATCHER(InitListExpr, isFullyDesignated) {
+ return std::all_of(Node.begin(), Node.end(), [](auto *InitExpr) {
+ return isa<DesignatedInitExpr>(InitExpr);
+ });
+}
+
+AST_MATCHER(InitListExpr, hasSingleElement) { return Node.getNumInits() == 1; }
+
+AST_MATCHER_FUNCTION(::internal::Matcher<CXXRecordDecl>, hasBaseWithFields) {
+ return hasAnyBase(hasType(cxxRecordDecl(has(fieldDecl()))));
+}
+
+AST_MATCHER(FieldDecl, isAnonymousDecl) {
+ if (const auto *Record =
+ Node.getType().getCanonicalType()->getAsRecordDecl()) {
+ return Record->isAnonymousStructOrUnion() || !Record->getIdentifier();
+ }
+ return false;
+}
+
+void UseDesignatedInitializersCheck::registerMatchers(MatchFinder *Finder) {
+ Finder->addMatcher(
+ initListExpr(
+ hasType(cxxRecordDecl(RestrictToPODTypes ? isPOD() : isAggregate(),
+ unless(hasBaseWithFields()),
+ unless(has(fieldDecl(isAnonymousDecl()))))
+ .bind("type")),
+ unless(IgnoreSingleElementAggregates ? hasSingleElement()
+ : unless(anything())),
+ unless(isFullyDesignated()))
+ .bind("init"),
+ this);
+}
+
+static bool isFullyUndesignated(const InitListExpr *SyntacticInitList) {
+ return std::all_of(
+ SyntacticInitList->begin(), SyntacticInitList->end(),
+ [](auto *InitExpr) { return !isa<DesignatedInitExpr>(InitExpr); });
+}
+
+void UseDesignatedInitializersCheck::check(
+ const MatchFinder::MatchResult &Result) {
+ const auto *InitList = Result.Nodes.getNodeAs<InitListExpr>("init");
+ const auto *Type = Result.Nodes.getNodeAs<CXXRecordDecl>("type");
+ if (!Type || !InitList)
+ return;
+ if (const auto *SyntacticInitList = InitList->getSyntacticForm()) {
+ const llvm::DenseMap<clang::SourceLocation, std::string> Designators =
+ clang::tooling::getDesignators(SyntacticInitList);
+ if (isFullyUndesignated(SyntacticInitList)) {
+ std::string NewList = "{";
+ for (const Stmt *InitExpr : *SyntacticInitList) {
+ if (InitExpr != *SyntacticInitList->begin())
+ NewList += ", ";
+ NewList += Designators.at(InitExpr->getBeginLoc());
+ NewList += "=";
----------------
SimplyDanny wrote:
I have no objections. However, this is the style that's also used by the inlay hints generated by Clangd which is the reason why I omitted the spaces. Does that change your opinion?
https://github.com/llvm/llvm-project/pull/80541
More information about the cfe-commits
mailing list