1 //===--- LoopConvertUtils.cpp - clang-tidy --------------------------------===// 2 // 3 // The LLVM Compiler Infrastructure 4 // 5 // This file is distributed under the University of Illinois Open Source 6 // License. See LICENSE.TXT for details. 7 // 8 //===----------------------------------------------------------------------===// 9 10 #include "LoopConvertUtils.h" 11 #include "clang/Basic/IdentifierTable.h" 12 #include "clang/Basic/LLVM.h" 13 #include "clang/Basic/Lambda.h" 14 #include "clang/Basic/SourceManager.h" 15 #include "clang/Basic/SourceLocation.h" 16 #include "clang/Basic/TokenKinds.h" 17 #include "clang/Lex/Lexer.h" 18 #include "llvm/ADT/APSInt.h" 19 #include "llvm/ADT/FoldingSet.h" 20 #include "llvm/ADT/StringRef.h" 21 #include "llvm/Support/Casting.h" 22 #include <algorithm> 23 #include <cassert> 24 #include <cstddef> 25 #include <string> 26 #include <utility> 27 28 using namespace clang::ast_matchers; 29 30 namespace clang { 31 namespace tidy { 32 namespace modernize { 33 34 /// \brief Tracks a stack of parent statements during traversal. 35 /// 36 /// All this really does is inject push_back() before running 37 /// RecursiveASTVisitor::TraverseStmt() and pop_back() afterwards. The Stmt atop 38 /// the stack is the parent of the current statement (NULL for the topmost 39 /// statement). 40 bool StmtAncestorASTVisitor::TraverseStmt(Stmt *Statement) { 41 StmtAncestors.insert(std::make_pair(Statement, StmtStack.back())); 42 StmtStack.push_back(Statement); 43 RecursiveASTVisitor<StmtAncestorASTVisitor>::TraverseStmt(Statement); 44 StmtStack.pop_back(); 45 return true; 46 } 47 48 /// \brief Keep track of the DeclStmt associated with each VarDecl. 49 /// 50 /// Combined with StmtAncestors, this provides roughly the same information as 51 /// Scope, as we can map a VarDecl to its DeclStmt, then walk up the parent tree 52 /// using StmtAncestors. 53 bool StmtAncestorASTVisitor::VisitDeclStmt(DeclStmt *Decls) { 54 for (const auto *decl : Decls->decls()) { 55 if (const auto *V = dyn_cast<VarDecl>(decl)) 56 DeclParents.insert(std::make_pair(V, Decls)); 57 } 58 return true; 59 } 60 61 /// \brief record the DeclRefExpr as part of the parent expression. 62 bool ComponentFinderASTVisitor::VisitDeclRefExpr(DeclRefExpr *E) { 63 Components.push_back(E); 64 return true; 65 } 66 67 /// \brief record the MemberExpr as part of the parent expression. 68 bool ComponentFinderASTVisitor::VisitMemberExpr(MemberExpr *Member) { 69 Components.push_back(Member); 70 return true; 71 } 72 73 /// \brief Forward any DeclRefExprs to a check on the referenced variable 74 /// declaration. 75 bool DependencyFinderASTVisitor::VisitDeclRefExpr(DeclRefExpr *DeclRef) { 76 if (auto *V = dyn_cast_or_null<VarDecl>(DeclRef->getDecl())) 77 return VisitVarDecl(V); 78 return true; 79 } 80 81 /// \brief Determine if any this variable is declared inside the ContainingStmt. 82 bool DependencyFinderASTVisitor::VisitVarDecl(VarDecl *V) { 83 const Stmt *Curr = DeclParents->lookup(V); 84 // First, see if the variable was declared within an inner scope of the loop. 85 while (Curr != nullptr) { 86 if (Curr == ContainingStmt) { 87 DependsOnInsideVariable = true; 88 return false; 89 } 90 Curr = StmtParents->lookup(Curr); 91 } 92 93 // Next, check if the variable was removed from existence by an earlier 94 // iteration. 95 for (const auto &I : *ReplacedVars) { 96 if (I.second == V) { 97 DependsOnInsideVariable = true; 98 return false; 99 } 100 } 101 return true; 102 } 103 104 /// \brief If we already created a variable for TheLoop, check to make sure 105 /// that the name was not already taken. 106 bool DeclFinderASTVisitor::VisitForStmt(ForStmt *TheLoop) { 107 StmtGeneratedVarNameMap::const_iterator I = GeneratedDecls->find(TheLoop); 108 if (I != GeneratedDecls->end() && I->second == Name) { 109 Found = true; 110 return false; 111 } 112 return true; 113 } 114 115 /// \brief If any named declaration within the AST subtree has the same name, 116 /// then consider Name already taken. 117 bool DeclFinderASTVisitor::VisitNamedDecl(NamedDecl *D) { 118 const IdentifierInfo *Ident = D->getIdentifier(); 119 if (Ident && Ident->getName() == Name) { 120 Found = true; 121 return false; 122 } 123 return true; 124 } 125 126 /// \brief Forward any declaration references to the actual check on the 127 /// referenced declaration. 128 bool DeclFinderASTVisitor::VisitDeclRefExpr(DeclRefExpr *DeclRef) { 129 if (auto *D = dyn_cast<NamedDecl>(DeclRef->getDecl())) 130 return VisitNamedDecl(D); 131 return true; 132 } 133 134 /// \brief If the new variable name conflicts with any type used in the loop, 135 /// then we mark that variable name as taken. 136 bool DeclFinderASTVisitor::VisitTypeLoc(TypeLoc TL) { 137 QualType QType = TL.getType(); 138 139 // Check if our name conflicts with a type, to handle for typedefs. 140 if (QType.getAsString() == Name) { 141 Found = true; 142 return false; 143 } 144 // Check for base type conflicts. For example, when a struct is being 145 // referenced in the body of the loop, the above getAsString() will return the 146 // whole type (ex. "struct s"), but will be caught here. 147 if (const IdentifierInfo *Ident = QType.getBaseTypeIdentifier()) { 148 if (Ident->getName() == Name) { 149 Found = true; 150 return false; 151 } 152 } 153 return true; 154 } 155 156 /// \brief Look through conversion/copy constructors to find the explicit 157 /// initialization expression, returning it is found. 158 /// 159 /// The main idea is that given 160 /// vector<int> v; 161 /// we consider either of these initializations 162 /// vector<int>::iterator it = v.begin(); 163 /// vector<int>::iterator it(v.begin()); 164 /// and retrieve `v.begin()` as the expression used to initialize `it` but do 165 /// not include 166 /// vector<int>::iterator it; 167 /// vector<int>::iterator it(v.begin(), 0); // if this constructor existed 168 /// as being initialized from `v.begin()` 169 const Expr *digThroughConstructors(const Expr *E) { 170 if (!E) 171 return nullptr; 172 E = E->IgnoreImplicit(); 173 if (const auto *ConstructExpr = dyn_cast<CXXConstructExpr>(E)) { 174 // The initial constructor must take exactly one parameter, but base class 175 // and deferred constructors can take more. 176 if (ConstructExpr->getNumArgs() != 1 || 177 ConstructExpr->getConstructionKind() != CXXConstructExpr::CK_Complete) 178 return nullptr; 179 E = ConstructExpr->getArg(0); 180 if (const auto *Temp = dyn_cast<MaterializeTemporaryExpr>(E)) 181 E = Temp->GetTemporaryExpr(); 182 return digThroughConstructors(E); 183 } 184 return E; 185 } 186 187 /// \brief Returns true when two Exprs are equivalent. 188 bool areSameExpr(ASTContext *Context, const Expr *First, const Expr *Second) { 189 if (!First || !Second) 190 return false; 191 192 llvm::FoldingSetNodeID FirstID, SecondID; 193 First->Profile(FirstID, *Context, true); 194 Second->Profile(SecondID, *Context, true); 195 return FirstID == SecondID; 196 } 197 198 /// \brief Returns the DeclRefExpr represented by E, or NULL if there isn't one. 199 const DeclRefExpr *getDeclRef(const Expr *E) { 200 return dyn_cast<DeclRefExpr>(E->IgnoreParenImpCasts()); 201 } 202 203 /// \brief Returns true when two ValueDecls are the same variable. 204 bool areSameVariable(const ValueDecl *First, const ValueDecl *Second) { 205 return First && Second && 206 First->getCanonicalDecl() == Second->getCanonicalDecl(); 207 } 208 209 /// \brief Determines if an expression is a declaration reference to a 210 /// particular variable. 211 static bool exprReferencesVariable(const ValueDecl *Target, const Expr *E) { 212 if (!Target || !E) 213 return false; 214 const DeclRefExpr *Decl = getDeclRef(E); 215 return Decl && areSameVariable(Target, Decl->getDecl()); 216 } 217 218 /// \brief If the expression is a dereference or call to operator*(), return the 219 /// operand. Otherwise, return NULL. 220 static const Expr *getDereferenceOperand(const Expr *E) { 221 if (const auto *Uop = dyn_cast<UnaryOperator>(E)) 222 return Uop->getOpcode() == UO_Deref ? Uop->getSubExpr() : nullptr; 223 224 if (const auto *OpCall = dyn_cast<CXXOperatorCallExpr>(E)) { 225 return OpCall->getOperator() == OO_Star && OpCall->getNumArgs() == 1 226 ? OpCall->getArg(0) 227 : nullptr; 228 } 229 230 return nullptr; 231 } 232 233 /// \brief Returns true when the Container contains an Expr equivalent to E. 234 template <typename ContainerT> 235 static bool containsExpr(ASTContext *Context, const ContainerT *Container, 236 const Expr *E) { 237 llvm::FoldingSetNodeID ID; 238 E->Profile(ID, *Context, true); 239 for (const auto &I : *Container) { 240 if (ID == I.second) 241 return true; 242 } 243 return false; 244 } 245 246 /// \brief Returns true when the index expression is a declaration reference to 247 /// IndexVar. 248 /// 249 /// If the index variable is `index`, this function returns true on 250 /// arrayExpression[index]; 251 /// containerExpression[index]; 252 /// but not 253 /// containerExpression[notIndex]; 254 static bool isIndexInSubscriptExpr(const Expr *IndexExpr, 255 const VarDecl *IndexVar) { 256 const DeclRefExpr *Idx = getDeclRef(IndexExpr); 257 return Idx && Idx->getType()->isIntegerType() && 258 areSameVariable(IndexVar, Idx->getDecl()); 259 } 260 261 /// \brief Returns true when the index expression is a declaration reference to 262 /// IndexVar, Obj is the same expression as SourceExpr after all parens and 263 /// implicit casts are stripped off. 264 /// 265 /// If PermitDeref is true, IndexExpression may 266 /// be a dereference (overloaded or builtin operator*). 267 /// 268 /// This function is intended for array-like containers, as it makes sure that 269 /// both the container and the index match. 270 /// If the loop has index variable `index` and iterates over `container`, then 271 /// isIndexInSubscriptExpr returns true for 272 /// \code 273 /// container[index] 274 /// container.at(index) 275 /// container->at(index) 276 /// \endcode 277 /// but not for 278 /// \code 279 /// container[notIndex] 280 /// notContainer[index] 281 /// \endcode 282 /// If PermitDeref is true, then isIndexInSubscriptExpr additionally returns 283 /// true on these expressions: 284 /// \code 285 /// (*container)[index] 286 /// (*container).at(index) 287 /// \endcode 288 static bool isIndexInSubscriptExpr(ASTContext *Context, const Expr *IndexExpr, 289 const VarDecl *IndexVar, const Expr *Obj, 290 const Expr *SourceExpr, bool PermitDeref) { 291 if (!SourceExpr || !Obj || !isIndexInSubscriptExpr(IndexExpr, IndexVar)) 292 return false; 293 294 if (areSameExpr(Context, SourceExpr->IgnoreParenImpCasts(), 295 Obj->IgnoreParenImpCasts())) 296 return true; 297 298 if (const Expr *InnerObj = getDereferenceOperand(Obj->IgnoreParenImpCasts())) 299 if (PermitDeref && areSameExpr(Context, SourceExpr->IgnoreParenImpCasts(), 300 InnerObj->IgnoreParenImpCasts())) 301 return true; 302 303 return false; 304 } 305 306 /// \brief Returns true when Opcall is a call a one-parameter dereference of 307 /// IndexVar. 308 /// 309 /// For example, if the index variable is `index`, returns true for 310 /// *index 311 /// but not 312 /// index 313 /// *notIndex 314 static bool isDereferenceOfOpCall(const CXXOperatorCallExpr *OpCall, 315 const VarDecl *IndexVar) { 316 return OpCall->getOperator() == OO_Star && OpCall->getNumArgs() == 1 && 317 exprReferencesVariable(IndexVar, OpCall->getArg(0)); 318 } 319 320 /// \brief Returns true when Uop is a dereference of IndexVar. 321 /// 322 /// For example, if the index variable is `index`, returns true for 323 /// *index 324 /// but not 325 /// index 326 /// *notIndex 327 static bool isDereferenceOfUop(const UnaryOperator *Uop, 328 const VarDecl *IndexVar) { 329 return Uop->getOpcode() == UO_Deref && 330 exprReferencesVariable(IndexVar, Uop->getSubExpr()); 331 } 332 333 /// \brief Determines whether the given Decl defines a variable initialized to 334 /// the loop object. 335 /// 336 /// This is intended to find cases such as 337 /// \code 338 /// for (int i = 0; i < arraySize(arr); ++i) { 339 /// T t = arr[i]; 340 /// // use t, do not use i 341 /// } 342 /// \endcode 343 /// and 344 /// \code 345 /// for (iterator i = container.begin(), e = container.end(); i != e; ++i) { 346 /// T t = *i; 347 /// // use t, do not use i 348 /// } 349 /// \endcode 350 static bool isAliasDecl(ASTContext *Context, const Decl *TheDecl, 351 const VarDecl *IndexVar) { 352 const auto *VDecl = dyn_cast<VarDecl>(TheDecl); 353 if (!VDecl) 354 return false; 355 if (!VDecl->hasInit()) 356 return false; 357 358 bool OnlyCasts = true; 359 const Expr *Init = VDecl->getInit()->IgnoreParenImpCasts(); 360 if (Init && isa<CXXConstructExpr>(Init)) { 361 Init = digThroughConstructors(Init); 362 OnlyCasts = false; 363 } 364 if (!Init) 365 return false; 366 367 // Check that the declared type is the same as (or a reference to) the 368 // container type. 369 if (!OnlyCasts) { 370 QualType InitType = Init->getType(); 371 QualType DeclarationType = VDecl->getType(); 372 if (!DeclarationType.isNull() && DeclarationType->isReferenceType()) 373 DeclarationType = DeclarationType.getNonReferenceType(); 374 375 if (InitType.isNull() || DeclarationType.isNull() || 376 !Context->hasSameUnqualifiedType(DeclarationType, InitType)) 377 return false; 378 } 379 380 switch (Init->getStmtClass()) { 381 case Stmt::ArraySubscriptExprClass: { 382 const auto *E = cast<ArraySubscriptExpr>(Init); 383 // We don't really care which array is used here. We check to make sure 384 // it was the correct one later, since the AST will traverse it next. 385 return isIndexInSubscriptExpr(E->getIdx(), IndexVar); 386 } 387 388 case Stmt::UnaryOperatorClass: 389 return isDereferenceOfUop(cast<UnaryOperator>(Init), IndexVar); 390 391 case Stmt::CXXOperatorCallExprClass: { 392 const auto *OpCall = cast<CXXOperatorCallExpr>(Init); 393 if (OpCall->getOperator() == OO_Star) 394 return isDereferenceOfOpCall(OpCall, IndexVar); 395 if (OpCall->getOperator() == OO_Subscript) { 396 assert(OpCall->getNumArgs() == 2); 397 return isIndexInSubscriptExpr(OpCall->getArg(1), IndexVar); 398 } 399 break; 400 } 401 402 case Stmt::CXXMemberCallExprClass: { 403 const auto *MemCall = cast<CXXMemberCallExpr>(Init); 404 // This check is needed because getMethodDecl can return nullptr if the 405 // callee is a member function pointer. 406 const auto *MDecl = MemCall->getMethodDecl(); 407 if (MDecl && !isa<CXXConversionDecl>(MDecl) && 408 MDecl->getNameAsString() == "at" && MemCall->getNumArgs() == 1) { 409 return isIndexInSubscriptExpr(MemCall->getArg(0), IndexVar); 410 } 411 return false; 412 } 413 414 default: 415 break; 416 } 417 return false; 418 } 419 420 /// \brief Determines whether the bound of a for loop condition expression is 421 /// the same as the statically computable size of ArrayType. 422 /// 423 /// Given 424 /// \code 425 /// const int N = 5; 426 /// int arr[N]; 427 /// \endcode 428 /// This is intended to permit 429 /// \code 430 /// for (int i = 0; i < N; ++i) { /* use arr[i] */ } 431 /// for (int i = 0; i < arraysize(arr); ++i) { /* use arr[i] */ } 432 /// \endcode 433 static bool arrayMatchesBoundExpr(ASTContext *Context, 434 const QualType &ArrayType, 435 const Expr *ConditionExpr) { 436 if (!ConditionExpr || ConditionExpr->isValueDependent()) 437 return false; 438 const ConstantArrayType *ConstType = 439 Context->getAsConstantArrayType(ArrayType); 440 if (!ConstType) 441 return false; 442 llvm::APSInt ConditionSize; 443 if (!ConditionExpr->isIntegerConstantExpr(ConditionSize, *Context)) 444 return false; 445 llvm::APSInt ArraySize(ConstType->getSize()); 446 return llvm::APSInt::isSameValue(ConditionSize, ArraySize); 447 } 448 449 ForLoopIndexUseVisitor::ForLoopIndexUseVisitor(ASTContext *Context, 450 const VarDecl *IndexVar, 451 const VarDecl *EndVar, 452 const Expr *ContainerExpr, 453 const Expr *ArrayBoundExpr, 454 bool ContainerNeedsDereference) 455 : Context(Context), IndexVar(IndexVar), EndVar(EndVar), 456 ContainerExpr(ContainerExpr), ArrayBoundExpr(ArrayBoundExpr), 457 ContainerNeedsDereference(ContainerNeedsDereference), 458 OnlyUsedAsIndex(true), AliasDecl(nullptr), 459 ConfidenceLevel(Confidence::CL_Safe), NextStmtParent(nullptr), 460 CurrStmtParent(nullptr), ReplaceWithAliasUse(false), 461 AliasFromForInit(false) { 462 if (ContainerExpr) 463 addComponent(ContainerExpr); 464 } 465 466 bool ForLoopIndexUseVisitor::findAndVerifyUsages(const Stmt *Body) { 467 TraverseStmt(const_cast<Stmt *>(Body)); 468 return OnlyUsedAsIndex && ContainerExpr; 469 } 470 471 void ForLoopIndexUseVisitor::addComponents(const ComponentVector &Components) { 472 // FIXME: add sort(on ID)+unique to avoid extra work. 473 for (const auto &I : Components) 474 addComponent(I); 475 } 476 477 void ForLoopIndexUseVisitor::addComponent(const Expr *E) { 478 llvm::FoldingSetNodeID ID; 479 const Expr *Node = E->IgnoreParenImpCasts(); 480 Node->Profile(ID, *Context, true); 481 DependentExprs.push_back(std::make_pair(Node, ID)); 482 } 483 484 void ForLoopIndexUseVisitor::addUsage(const Usage &U) { 485 SourceLocation Begin = U.Range.getBegin(); 486 if (Begin.isMacroID()) 487 Begin = Context->getSourceManager().getSpellingLoc(Begin); 488 489 if (UsageLocations.insert(Begin).second) 490 Usages.push_back(U); 491 } 492 493 /// \brief If the unary operator is a dereference of IndexVar, include it 494 /// as a valid usage and prune the traversal. 495 /// 496 /// For example, if container.begin() and container.end() both return pointers 497 /// to int, this makes sure that the initialization for `k` is not counted as an 498 /// unconvertible use of the iterator `i`. 499 /// \code 500 /// for (int *i = container.begin(), *e = container.end(); i != e; ++i) { 501 /// int k = *i + 2; 502 /// } 503 /// \endcode 504 bool ForLoopIndexUseVisitor::TraverseUnaryDeref(UnaryOperator *Uop) { 505 // If we dereference an iterator that's actually a pointer, count the 506 // occurrence. 507 if (isDereferenceOfUop(Uop, IndexVar)) { 508 addUsage(Usage(Uop)); 509 return true; 510 } 511 512 return VisitorBase::TraverseUnaryOperator(Uop); 513 } 514 515 /// \brief If the member expression is operator-> (overloaded or not) on 516 /// IndexVar, include it as a valid usage and prune the traversal. 517 /// 518 /// For example, given 519 /// \code 520 /// struct Foo { int bar(); int x; }; 521 /// vector<Foo> v; 522 /// \endcode 523 /// the following uses will be considered convertible: 524 /// \code 525 /// for (vector<Foo>::iterator i = v.begin(), e = v.end(); i != e; ++i) { 526 /// int b = i->bar(); 527 /// int k = i->x + 1; 528 /// } 529 /// \endcode 530 /// though 531 /// \code 532 /// for (vector<Foo>::iterator i = v.begin(), e = v.end(); i != e; ++i) { 533 /// int k = i.insert(1); 534 /// } 535 /// for (vector<Foo>::iterator i = v.begin(), e = v.end(); i != e; ++i) { 536 /// int b = e->bar(); 537 /// } 538 /// \endcode 539 /// will not. 540 bool ForLoopIndexUseVisitor::TraverseMemberExpr(MemberExpr *Member) { 541 const Expr *Base = Member->getBase(); 542 const DeclRefExpr *Obj = getDeclRef(Base); 543 const Expr *ResultExpr = Member; 544 QualType ExprType; 545 if (const auto *Call = 546 dyn_cast<CXXOperatorCallExpr>(Base->IgnoreParenImpCasts())) { 547 // If operator->() is a MemberExpr containing a CXXOperatorCallExpr, then 548 // the MemberExpr does not have the expression we want. We therefore catch 549 // that instance here. 550 // For example, if vector<Foo>::iterator defines operator->(), then the 551 // example `i->bar()` at the top of this function is a CXXMemberCallExpr 552 // referring to `i->` as the member function called. We want just `i`, so 553 // we take the argument to operator->() as the base object. 554 if (Call->getOperator() == OO_Arrow) { 555 assert(Call->getNumArgs() == 1 && 556 "Operator-> takes more than one argument"); 557 Obj = getDeclRef(Call->getArg(0)); 558 ResultExpr = Obj; 559 ExprType = Call->getCallReturnType(*Context); 560 } 561 } 562 563 if (Obj && exprReferencesVariable(IndexVar, Obj)) { 564 // Member calls on the iterator with '.' are not allowed. 565 if (!Member->isArrow()) { 566 OnlyUsedAsIndex = false; 567 return true; 568 } 569 570 if (ExprType.isNull()) 571 ExprType = Obj->getType(); 572 573 if (!ExprType->isPointerType()) 574 return false; 575 576 // FIXME: This works around not having the location of the arrow operator. 577 // Consider adding OperatorLoc to MemberExpr? 578 SourceLocation ArrowLoc = Lexer::getLocForEndOfToken( 579 Base->getExprLoc(), 0, Context->getSourceManager(), 580 Context->getLangOpts()); 581 // If something complicated is happening (i.e. the next token isn't an 582 // arrow), give up on making this work. 583 if (ArrowLoc.isValid()) { 584 addUsage(Usage(ResultExpr, Usage::UK_MemberThroughArrow, 585 SourceRange(Base->getExprLoc(), ArrowLoc))); 586 return true; 587 } 588 } 589 return VisitorBase::TraverseMemberExpr(Member); 590 } 591 592 /// \brief If a member function call is the at() accessor on the container with 593 /// IndexVar as the single argument, include it as a valid usage and prune 594 /// the traversal. 595 /// 596 /// Member calls on other objects will not be permitted. 597 /// Calls on the iterator object are not permitted, unless done through 598 /// operator->(). The one exception is allowing vector::at() for pseudoarrays. 599 bool ForLoopIndexUseVisitor::TraverseCXXMemberCallExpr( 600 CXXMemberCallExpr *MemberCall) { 601 auto *Member = 602 dyn_cast<MemberExpr>(MemberCall->getCallee()->IgnoreParenImpCasts()); 603 if (!Member) 604 return VisitorBase::TraverseCXXMemberCallExpr(MemberCall); 605 606 // We specifically allow an accessor named "at" to let STL in, though 607 // this is restricted to pseudo-arrays by requiring a single, integer 608 // argument. 609 const IdentifierInfo *Ident = Member->getMemberDecl()->getIdentifier(); 610 if (Ident && Ident->isStr("at") && MemberCall->getNumArgs() == 1) { 611 if (isIndexInSubscriptExpr(Context, MemberCall->getArg(0), IndexVar, 612 Member->getBase(), ContainerExpr, 613 ContainerNeedsDereference)) { 614 addUsage(Usage(MemberCall)); 615 return true; 616 } 617 } 618 619 if (containsExpr(Context, &DependentExprs, Member->getBase())) 620 ConfidenceLevel.lowerTo(Confidence::CL_Risky); 621 622 return VisitorBase::TraverseCXXMemberCallExpr(MemberCall); 623 } 624 625 /// \brief If an overloaded operator call is a dereference of IndexVar or 626 /// a subscript of the container with IndexVar as the single argument, 627 /// include it as a valid usage and prune the traversal. 628 /// 629 /// For example, given 630 /// \code 631 /// struct Foo { int bar(); int x; }; 632 /// vector<Foo> v; 633 /// void f(Foo); 634 /// \endcode 635 /// the following uses will be considered convertible: 636 /// \code 637 /// for (vector<Foo>::iterator i = v.begin(), e = v.end(); i != e; ++i) { 638 /// f(*i); 639 /// } 640 /// for (int i = 0; i < v.size(); ++i) { 641 /// int i = v[i] + 1; 642 /// } 643 /// \endcode 644 bool ForLoopIndexUseVisitor::TraverseCXXOperatorCallExpr( 645 CXXOperatorCallExpr *OpCall) { 646 switch (OpCall->getOperator()) { 647 case OO_Star: 648 if (isDereferenceOfOpCall(OpCall, IndexVar)) { 649 addUsage(Usage(OpCall)); 650 return true; 651 } 652 break; 653 654 case OO_Subscript: 655 if (OpCall->getNumArgs() != 2) 656 break; 657 if (isIndexInSubscriptExpr(Context, OpCall->getArg(1), IndexVar, 658 OpCall->getArg(0), ContainerExpr, 659 ContainerNeedsDereference)) { 660 addUsage(Usage(OpCall)); 661 return true; 662 } 663 break; 664 665 default: 666 break; 667 } 668 return VisitorBase::TraverseCXXOperatorCallExpr(OpCall); 669 } 670 671 /// \brief If we encounter an array with IndexVar as the index of an 672 /// ArraySubsriptExpression, note it as a consistent usage and prune the 673 /// AST traversal. 674 /// 675 /// For example, given 676 /// \code 677 /// const int N = 5; 678 /// int arr[N]; 679 /// \endcode 680 /// This is intended to permit 681 /// \code 682 /// for (int i = 0; i < N; ++i) { /* use arr[i] */ } 683 /// \endcode 684 /// but not 685 /// \code 686 /// for (int i = 0; i < N; ++i) { /* use notArr[i] */ } 687 /// \endcode 688 /// and further checking needs to be done later to ensure that exactly one array 689 /// is referenced. 690 bool ForLoopIndexUseVisitor::TraverseArraySubscriptExpr(ArraySubscriptExpr *E) { 691 Expr *Arr = E->getBase(); 692 if (!isIndexInSubscriptExpr(E->getIdx(), IndexVar)) 693 return VisitorBase::TraverseArraySubscriptExpr(E); 694 695 if ((ContainerExpr && 696 !areSameExpr(Context, Arr->IgnoreParenImpCasts(), 697 ContainerExpr->IgnoreParenImpCasts())) || 698 !arrayMatchesBoundExpr(Context, Arr->IgnoreImpCasts()->getType(), 699 ArrayBoundExpr)) { 700 // If we have already discovered the array being indexed and this isn't it 701 // or this array doesn't match, mark this loop as unconvertible. 702 OnlyUsedAsIndex = false; 703 return VisitorBase::TraverseArraySubscriptExpr(E); 704 } 705 706 if (!ContainerExpr) 707 ContainerExpr = Arr; 708 709 addUsage(Usage(E)); 710 return true; 711 } 712 713 /// \brief If we encounter a reference to IndexVar in an unpruned branch of the 714 /// traversal, mark this loop as unconvertible. 715 /// 716 /// This implements the whitelist for convertible loops: any usages of IndexVar 717 /// not explicitly considered convertible by this traversal will be caught by 718 /// this function. 719 /// 720 /// Additionally, if the container expression is more complex than just a 721 /// DeclRefExpr, and some part of it is appears elsewhere in the loop, lower 722 /// our confidence in the transformation. 723 /// 724 /// For example, these are not permitted: 725 /// \code 726 /// for (int i = 0; i < N; ++i) { printf("arr[%d] = %d", i, arr[i]); } 727 /// for (vector<int>::iterator i = container.begin(), e = container.end(); 728 /// i != e; ++i) 729 /// i.insert(0); 730 /// for (vector<int>::iterator i = container.begin(), e = container.end(); 731 /// i != e; ++i) 732 /// if (i + 1 != e) 733 /// printf("%d", *i); 734 /// \endcode 735 /// 736 /// And these will raise the risk level: 737 /// \code 738 /// int arr[10][20]; 739 /// int l = 5; 740 /// for (int j = 0; j < 20; ++j) 741 /// int k = arr[l][j] + l; // using l outside arr[l] is considered risky 742 /// for (int i = 0; i < obj.getVector().size(); ++i) 743 /// obj.foo(10); // using `obj` is considered risky 744 /// \endcode 745 bool ForLoopIndexUseVisitor::VisitDeclRefExpr(DeclRefExpr *E) { 746 const ValueDecl *TheDecl = E->getDecl(); 747 if (areSameVariable(IndexVar, TheDecl) || 748 exprReferencesVariable(IndexVar, E) || areSameVariable(EndVar, TheDecl) || 749 exprReferencesVariable(EndVar, E)) 750 OnlyUsedAsIndex = false; 751 if (containsExpr(Context, &DependentExprs, E)) 752 ConfidenceLevel.lowerTo(Confidence::CL_Risky); 753 return true; 754 } 755 756 /// \brief If the loop index is captured by a lambda, replace this capture 757 /// by the range-for loop variable. 758 /// 759 /// For example: 760 /// \code 761 /// for (int i = 0; i < N; ++i) { 762 /// auto f = [v, i](int k) { 763 /// printf("%d\n", v[i] + k); 764 /// }; 765 /// f(v[i]); 766 /// } 767 /// \endcode 768 /// 769 /// Will be replaced by: 770 /// \code 771 /// for (auto & elem : v) { 772 /// auto f = [v, elem](int k) { 773 /// printf("%d\n", elem + k); 774 /// }; 775 /// f(elem); 776 /// } 777 /// \endcode 778 bool ForLoopIndexUseVisitor::TraverseLambdaCapture(LambdaExpr *LE, 779 const LambdaCapture *C, 780 Expr *Init) { 781 if (C->capturesVariable()) { 782 const VarDecl *VDecl = C->getCapturedVar(); 783 if (areSameVariable(IndexVar, cast<ValueDecl>(VDecl))) { 784 // FIXME: if the index is captured, it will count as an usage and the 785 // alias (if any) won't work, because it is only used in case of having 786 // exactly one usage. 787 addUsage(Usage(nullptr, 788 C->getCaptureKind() == LCK_ByCopy ? Usage::UK_CaptureByCopy 789 : Usage::UK_CaptureByRef, 790 C->getLocation())); 791 } 792 } 793 return VisitorBase::TraverseLambdaCapture(LE, C, Init); 794 } 795 796 /// \brief If we find that another variable is created just to refer to the loop 797 /// element, note it for reuse as the loop variable. 798 /// 799 /// See the comments for isAliasDecl. 800 bool ForLoopIndexUseVisitor::VisitDeclStmt(DeclStmt *S) { 801 if (!AliasDecl && S->isSingleDecl() && 802 isAliasDecl(Context, S->getSingleDecl(), IndexVar)) { 803 AliasDecl = S; 804 if (CurrStmtParent) { 805 if (isa<IfStmt>(CurrStmtParent) || isa<WhileStmt>(CurrStmtParent) || 806 isa<SwitchStmt>(CurrStmtParent)) 807 ReplaceWithAliasUse = true; 808 else if (isa<ForStmt>(CurrStmtParent)) { 809 if (cast<ForStmt>(CurrStmtParent)->getConditionVariableDeclStmt() == S) 810 ReplaceWithAliasUse = true; 811 else 812 // It's assumed S came the for loop's init clause. 813 AliasFromForInit = true; 814 } 815 } 816 } 817 818 return true; 819 } 820 821 bool ForLoopIndexUseVisitor::TraverseStmt(Stmt *S) { 822 // If this is an initialization expression for a lambda capture, prune the 823 // traversal so that we don't end up diagnosing the contained DeclRefExpr as 824 // inconsistent usage. No need to record the usage here -- this is done in 825 // TraverseLambdaCapture(). 826 if (const auto *LE = dyn_cast_or_null<LambdaExpr>(NextStmtParent)) { 827 // Any child of a LambdaExpr that isn't the body is an initialization 828 // expression. 829 if (S != LE->getBody()) { 830 return true; 831 } 832 } 833 834 // All this pointer swapping is a mechanism for tracking immediate parentage 835 // of Stmts. 836 const Stmt *OldNextParent = NextStmtParent; 837 CurrStmtParent = NextStmtParent; 838 NextStmtParent = S; 839 bool Result = VisitorBase::TraverseStmt(S); 840 NextStmtParent = OldNextParent; 841 return Result; 842 } 843 844 std::string VariableNamer::createIndexName() { 845 // FIXME: Add in naming conventions to handle: 846 // - How to handle conflicts. 847 // - An interactive process for naming. 848 std::string IteratorName; 849 StringRef ContainerName; 850 if (TheContainer) 851 ContainerName = TheContainer->getName(); 852 853 size_t Len = ContainerName.size(); 854 if (Len > 1 && ContainerName.endswith(Style == NS_UpperCase ? "S" : "s")) { 855 IteratorName = ContainerName.substr(0, Len - 1); 856 // E.g.: (auto thing : things) 857 if (!declarationExists(IteratorName) || IteratorName == OldIndex->getName()) 858 return IteratorName; 859 } 860 861 if (Len > 2 && ContainerName.endswith(Style == NS_UpperCase ? "S_" : "s_")) { 862 IteratorName = ContainerName.substr(0, Len - 2); 863 // E.g.: (auto thing : things_) 864 if (!declarationExists(IteratorName) || IteratorName == OldIndex->getName()) 865 return IteratorName; 866 } 867 868 return OldIndex->getName(); 869 } 870 871 /// \brief Determines whether or not the the name \a Symbol conflicts with 872 /// language keywords or defined macros. Also checks if the name exists in 873 /// LoopContext, any of its parent contexts, or any of its child statements. 874 /// 875 /// We also check to see if the same identifier was generated by this loop 876 /// converter in a loop nested within SourceStmt. 877 bool VariableNamer::declarationExists(StringRef Symbol) { 878 assert(Context != nullptr && "Expected an ASTContext"); 879 IdentifierInfo &Ident = Context->Idents.get(Symbol); 880 881 // Check if the symbol is not an identifier (ie. is a keyword or alias). 882 if (!isAnyIdentifier(Ident.getTokenID())) 883 return true; 884 885 // Check for conflicting macro definitions. 886 if (Ident.hasMacroDefinition()) 887 return true; 888 889 // Determine if the symbol was generated in a parent context. 890 for (const Stmt *S = SourceStmt; S != nullptr; S = ReverseAST->lookup(S)) { 891 StmtGeneratedVarNameMap::const_iterator I = GeneratedDecls->find(S); 892 if (I != GeneratedDecls->end() && I->second == Symbol) 893 return true; 894 } 895 896 // FIXME: Rather than detecting conflicts at their usages, we should check the 897 // parent context. 898 // For some reason, lookup() always returns the pair (NULL, NULL) because its 899 // StoredDeclsMap is not initialized (i.e. LookupPtr.getInt() is false inside 900 // of DeclContext::lookup()). Why is this? 901 902 // Finally, determine if the symbol was used in the loop or a child context. 903 DeclFinderASTVisitor DeclFinder(Symbol, GeneratedDecls); 904 return DeclFinder.findUsages(SourceStmt); 905 } 906 907 } // namespace modernize 908 } // namespace tidy 909 } // namespace clang 910