On 08/14/2010 11:50 PM, Sean Hunt wrote:
Hey all,
This patch adds a new UDStringLiteral expression and the lexing and
parsing machinery required in order to implement this C++0x feature.
Yay!
Sean
Oops, my apologies, that was a reverse patch. Here's a corrected one.
diff --git a/include/clang/AST/ExprCXX.h b/include/clang/AST/ExprCXX.h
index 184edcd..8171ed5 100644
--- a/include/clang/AST/ExprCXX.h
+++ b/include/clang/AST/ExprCXX.h
@@ -2375,6 +2375,43 @@ public:
virtual child_iterator child_end();
};
+/// UDStringLiteralExpr - An expression for a user-defined
+/// string literal (e.g. "foo"_bar)
+///
+/// Both the DeclRefExpr and the IntegerConstant are fictional expressions
+/// generated from the literal.
+class UDStringLiteral : public CallExpr {
+ StringLiteral *BaseLiteral;
+
+public:
+ UDStringLiteral(ASTContext &C, StringLiteral *SL, Expr *fn, Expr **args,
+ unsigned numargs, QualType t)
+ : CallExpr(C, UDStringLiteralClass, fn, args, numargs, t, SourceLocation())
+ , BaseLiteral(SL) {
+ assert(numargs == 2 && "Wrong number of arguments to UDStringLiteral");
+ }
+
+ FunctionDecl *getLiteralOperator() { return getDirectCallee(); };
+ const FunctionDecl *getLiteralOperator() const { return getDirectCallee(); };
+
+ StringLiteral *getBaseLiteral() { return BaseLiteral; }
+ const StringLiteral *getBaseLiteral() const { return BaseLiteral; }
+ void setBaseLiteral(StringLiteral *SL) { BaseLiteral = SL; }
+
+ IdentifierInfo *getUDSuffix() const {
+ return getLiteralOperator()->getDeclName().getCXXLiteralIdentifier();
+ }
+
+ virtual SourceRange getSourceRange() const {
+ return getBaseLiteral()->getSourceRange();
+ }
+
+ static bool classof(const Stmt *T) {
+ return T->getStmtClass() == UDStringLiteralClass;
+ }
+ static bool classof(const UDStringLiteral *) { return true; }
+};
+
inline ExplicitTemplateArgumentList &OverloadExpr::getExplicitTemplateArgs() {
if (isa<UnresolvedLookupExpr>(this))
return cast<UnresolvedLookupExpr>(this)->getExplicitTemplateArgs();
diff --git a/include/clang/AST/RecursiveASTVisitor.h b/include/clang/AST/RecursiveASTVisitor.h
index ee0bca5..e12834a 100644
--- a/include/clang/AST/RecursiveASTVisitor.h
+++ b/include/clang/AST/RecursiveASTVisitor.h
@@ -1709,6 +1709,7 @@ DEF_TRAVERSE_STMT(FloatingLiteral, { })
DEF_TRAVERSE_STMT(ImaginaryLiteral, { })
DEF_TRAVERSE_STMT(StringLiteral, { })
DEF_TRAVERSE_STMT(ObjCStringLiteral, { })
+DEF_TRAVERSE_STMT(UDStringLiteral, { })
// FIXME: look at the following tricky-seeming exprs to see if we
// need to recurse on anything. These are ones that have methods
diff --git a/include/clang/Basic/DiagnosticLexKinds.td b/include/clang/Basic/DiagnosticLexKinds.td
index 9b06fa8..d1505de 100644
--- a/include/clang/Basic/DiagnosticLexKinds.td
+++ b/include/clang/Basic/DiagnosticLexKinds.td
@@ -98,6 +98,9 @@ def ext_string_too_long : Extension<"string literal of length %0 exceeds "
"maximum length %1 that %select{C90|ISO C99|C++}2 compilers are required to "
"support">, InGroup<OverlengthStrings>;
+def err_ud_suffix_mismatch : Error<"User-defined literal suffixes on adjacent "
+ "string literal tokens do not match">;
+
//===----------------------------------------------------------------------===//
// PTH Diagnostics
//===----------------------------------------------------------------------===//
diff --git a/include/clang/Basic/DiagnosticSemaKinds.td b/include/clang/Basic/DiagnosticSemaKinds.td
index 7cdc9f4..e056454 100644
--- a/include/clang/Basic/DiagnosticSemaKinds.td
+++ b/include/clang/Basic/DiagnosticSemaKinds.td
@@ -2928,9 +2928,13 @@ def err_operator_delete_param_type : Error<
// C++ literal operators
def err_literal_operator_outside_namespace : Error<
"literal operator %0 must be in a namespace or global scope">;
-// FIXME: This diagnostic sucks
+def warn_literal_operator_no_underscore : Warning< "literal operator names not "
+ "beginning with underscores are reserved for future standardization">;
+def err_literal_operator_overload : Error<
+ "no matching literal operator function for user-defined suffix '%0'">;
+// FIXME: These diagnostics suck
def err_literal_operator_params : Error<
- "parameter declaration for literal operator %0 is not valid">;
+ "parameter declaration for literal operator '%0' is not valid">;
// C++ conversion functions
def err_conv_function_not_member : Error<
diff --git a/include/clang/Basic/StmtNodes.td b/include/clang/Basic/StmtNodes.td
index a2f6973..20f75e7 100644
--- a/include/clang/Basic/StmtNodes.td
+++ b/include/clang/Basic/StmtNodes.td
@@ -112,6 +112,9 @@ def OverloadExpr : DStmt<Expr, 1>;
def UnresolvedLookupExpr : DStmt<OverloadExpr>;
def UnresolvedMemberExpr : DStmt<OverloadExpr>;
+// C++0x expressions
+def UDStringLiteral : DStmt<Expr>;
+
// Obj-C Expressions.
def ObjCStringLiteral : DStmt<Expr>;
def ObjCEncodeExpr : DStmt<Expr>;
diff --git a/include/clang/Lex/Lexer.h b/include/clang/Lex/Lexer.h
index 9e0fb7e..e05113d 100644
--- a/include/clang/Lex/Lexer.h
+++ b/include/clang/Lex/Lexer.h
@@ -17,6 +17,7 @@
#include "clang/Lex/PreprocessorLexer.h"
#include "clang/Basic/LangOptions.h"
#include "llvm/ADT/SmallVector.h"
+#include "llvm/Support/Allocator.h"
#include <string>
#include <vector>
#include <cassert>
@@ -67,6 +68,9 @@ class Lexer : public PreprocessorLexer {
// line" flag set on it.
bool IsAtStartOfLine;
+ // ExtraDataAllocator - An allocator for extra data on a token.
+ llvm::BumpPtrAllocator ExtraDataAllocator;
+
Lexer(const Lexer&); // DO NOT IMPLEMENT
void operator=(const Lexer&); // DO NOT IMPLEMENT
friend class Preprocessor;
diff --git a/include/clang/Lex/LiteralSupport.h b/include/clang/Lex/LiteralSupport.h
index ba46fb1..40112f5 100644
--- a/include/clang/Lex/LiteralSupport.h
+++ b/include/clang/Lex/LiteralSupport.h
@@ -27,6 +27,7 @@ class Preprocessor;
class Token;
class SourceLocation;
class TargetInfo;
+class IdentifierInfo;
/// NumericLiteralParser - This performs strict semantic analysis of the content
/// of a ppnumber, classifying it as either integer, floating, or erroneous,
@@ -145,6 +146,7 @@ class StringLiteralParser {
unsigned wchar_tByteWidth;
llvm::SmallString<512> ResultBuf;
char *ResultPtr; // cursor
+ IdentifierInfo *UDSuffix;
public:
StringLiteralParser(const Token *StringToks, unsigned NumStringToks,
Preprocessor &PP, bool Complain = true);
@@ -155,6 +157,9 @@ public:
const char *GetString() { return &ResultBuf[0]; }
unsigned GetStringLength() const { return ResultPtr-&ResultBuf[0]; }
+ bool isUserDefinedLiteral() const { return UDSuffix; }
+ IdentifierInfo *getUDSuffix() const { return UDSuffix; }
+
unsigned GetNumStringChars() const {
if (AnyWide)
return GetStringLength() / wchar_tByteWidth;
diff --git a/include/clang/Lex/Preprocessor.h b/include/clang/Lex/Preprocessor.h
index 6fd1943..ad4d372 100644
--- a/include/clang/Lex/Preprocessor.h
+++ b/include/clang/Lex/Preprocessor.h
@@ -616,7 +616,7 @@ public:
/// copy). The caller is not allowed to modify the returned buffer pointer
/// if an internal buffer is returned.
unsigned getSpelling(const Token &Tok, const char *&Buffer,
- bool *Invalid = 0) const;
+ bool *Invalid = 0, bool LiteralOnly = false) const;
/// getSpelling - This method is used to get the spelling of a token into a
/// SmallVector. Note that the returned StringRef may not point to the
diff --git a/include/clang/Lex/Token.h b/include/clang/Lex/Token.h
index bd9b468..18fa9bf 100644
--- a/include/clang/Lex/Token.h
+++ b/include/clang/Lex/Token.h
@@ -14,16 +14,16 @@
#ifndef LLVM_CLANG_TOKEN_H
#define LLVM_CLANG_TOKEN_H
+#include "llvm/Support/Allocator.h"
#include "clang/Basic/TemplateKinds.h"
#include "clang/Basic/TokenKinds.h"
#include "clang/Basic/SourceLocation.h"
#include "clang/Basic/OperatorKinds.h"
+#include "clang/Basic/IdentifierTable.h"
#include <cstdlib>
namespace clang {
-class IdentifierInfo;
-
/// Token - This structure provides full information about a lexed token.
/// It is not intended to be space efficient, it is intended to return as much
/// information as possible about each returned token. This is expected to be
@@ -34,6 +34,14 @@ class IdentifierInfo;
/// can be represented by a single typename annotation token that carries
/// information about the SourceRange of the tokens and the type object.
class Token {
+ /// An extra-large structure for storing the data needed for a user-defined
+ /// literal - the raw literal, and the identifier suffix.
+ struct UDLData {
+ IdentifierInfo *II;
+ const char *LiteralData;
+ unsigned LiteralLength;
+ };
+
/// The location of the token.
SourceLocation Loc;
@@ -47,7 +55,7 @@ class Token {
/// token.
unsigned UintData;
- /// PtrData - This is a union of four different pointer types, which depends
+ /// PtrData - This is a union of five different pointer types, which depends
/// on what type of token this is:
/// Identifiers, keywords, etc:
/// This is an IdentifierInfo*, which contains the uniqued identifier
@@ -55,6 +63,8 @@ class Token {
/// Literals: isLiteral() returns true.
/// This is a pointer to the start of the token in a text buffer, which
/// may be dirty (have trigraphs / escaped newlines).
+ /// User-defined literals: isUserDefinedLiteral() returns true.
+ /// This is a pointer to a UDLData.
/// Annotations (resolved type names, C++ scopes, etc): isAnnotation().
/// This is a pointer to sema-specific data for the annotation token.
/// Other:
@@ -71,12 +81,14 @@ class Token {
unsigned char Flags;
public:
- // Various flags set per token:
+ /// Various flags set per token:
enum TokenFlags {
- StartOfLine = 0x01, // At start of line or only after whitespace.
- LeadingSpace = 0x02, // Whitespace exists before this token.
- DisableExpand = 0x04, // This identifier may never be macro expanded.
- NeedsCleaning = 0x08 // Contained an escaped newline or trigraph.
+ StartOfLine = 0x01, ///< At start of line or only after whitespace
+ LeadingSpace = 0x02, ///< Whitespace exists before this token
+ DisableExpand = 0x04, ///< This identifier may never be macro expanded
+ NeedsCleaning = 0x08, ///< Contained an escaped newline or trigraph
+ UserDefinedLiteral = 0x10, ///< This literal has a ud-suffix
+ SLPortionClean = 0x20 ///< A UDL's literal portion needs no cleaning
};
tok::TokenKind getKind() const { return (tok::TokenKind)Kind; }
@@ -108,12 +120,40 @@ public:
assert(!isAnnotation() && "Annotation tokens have no length field");
return UintData;
}
+ /// getLiteralLength - Return the length of the literal portion of the token,
+ /// which may not be the token length if this is a user-defined literal.
+ unsigned getLiteralLength() const {
+ assert(isLiteral() && "Using getLiteralLength on a non-literal token");
+ if (isUserDefinedLiteral())
+ return reinterpret_cast<UDLData*>(PtrData)->LiteralLength;
+ else
+ return UintData;
+ }
void setLocation(SourceLocation L) { Loc = L; }
void setLength(unsigned Len) {
assert(!isAnnotation() && "Annotation tokens have no length field");
UintData = Len;
}
+ void setLiteralLength(unsigned Len) {
+ assert(isLiteral() && "Using setLiteralLength on a non-literal token");
+ if (isUserDefinedLiteral())
+ reinterpret_cast<UDLData*>(PtrData)->LiteralLength = Len;
+ else
+ UintData = Len;
+ }
+
+ /// makeUserDefinedLiteral - Set this token to be a user-defined literal
+ void makeUserDefinedLiteral(llvm::BumpPtrAllocator &Alloc) {
+ PtrData = Alloc.Allocate(sizeof(UDLData), 4);
+ setFlag(UserDefinedLiteral);
+ }
+ /// clearUserLiteralData - Deallocate the user-defined literal data
+ /// The token will be left in a garbage state.
+ void clearUserDefinedLiteral() {
+ delete reinterpret_cast<UDLData*>(PtrData);
+ PtrData = 0;
+ }
SourceLocation getAnnotationEndLoc() const {
assert(isAnnotation() && "Used AnnotEndLocID on non-annotation token");
@@ -154,11 +194,18 @@ public:
IdentifierInfo *getIdentifierInfo() const {
assert(!isAnnotation() && "Used IdentInfo on annotation token!");
- if (isLiteral()) return 0;
- return (IdentifierInfo*) PtrData;
+ if (isUserDefinedLiteral())
+ return reinterpret_cast<UDLData*>(PtrData)->II;
+ else if (isLiteral())
+ return 0;
+ else
+ return reinterpret_cast<IdentifierInfo*>(PtrData);
}
void setIdentifierInfo(IdentifierInfo *II) {
- PtrData = (void*) II;
+ if (isUserDefinedLiteral())
+ reinterpret_cast<UDLData*>(PtrData)->II = II;
+ else
+ PtrData = (void*)II;
}
/// getLiteralData - For a literal token (numeric constant, string, etc), this
@@ -166,11 +213,17 @@ public:
/// otherwise.
const char *getLiteralData() const {
assert(isLiteral() && "Cannot get literal data of non-literal");
- return reinterpret_cast<const char*>(PtrData);
+ if (isUserDefinedLiteral())
+ return reinterpret_cast<UDLData*>(PtrData)->LiteralData;
+ else
+ return reinterpret_cast<const char*>(PtrData);
}
void setLiteralData(const char *Ptr) {
assert(isLiteral() && "Cannot set literal data of non-literal");
- PtrData = const_cast<char*>(Ptr);
+ if (isUserDefinedLiteral())
+ reinterpret_cast<UDLData*>(PtrData)->LiteralData = Ptr;
+ else
+ PtrData = const_cast<char*>(Ptr);
}
void *getAnnotationValue() const {
@@ -221,6 +274,12 @@ public:
return (Flags & DisableExpand) ? true : false;
}
+ /// isUserDefinedLiteral - Return true if this is a C++0x user-defined literal
+ /// token.
+ bool isUserDefinedLiteral() const {
+ return (Flags & UserDefinedLiteral) ? true : false;
+ }
+
/// isObjCAtKeyword - Return true if we have an ObjC keyword identifier.
bool isObjCAtKeyword(tok::ObjCKeywordKind objcKey) const;
@@ -229,8 +288,17 @@ public:
/// needsCleaning - Return true if this token has trigraphs or escaped
/// newlines in it.
- ///
- bool needsCleaning() const { return (Flags & NeedsCleaning) ? true : false; }
+ bool needsCleaning() const {
+ return (Flags & NeedsCleaning) ? true : false;
+ }
+
+ /// literalNeedsCleaning - Return true if the literal portion of this token
+ /// needs cleaning.
+ bool literalNeedsCleaning() const {
+ assert(isLiteral() && "Using literalNeedsCleaning on a non-literal token");
+ return (Flags & NeedsCleaning) ? ((Flags & SLPortionClean) ? false : true)
+ : false;
+ }
};
/// PPConditionalInfo - Information about the conditional stack (#if directives)
diff --git a/include/clang/Parse/Action.h b/include/clang/Parse/Action.h
index 263dacb..c96461c 100644
--- a/include/clang/Parse/Action.h
+++ b/include/clang/Parse/Action.h
@@ -1123,7 +1123,7 @@ public:
/// ActOnStringLiteral - The specified tokens were lexed as pasted string
/// fragments (e.g. "foo" "bar" L"baz").
- virtual OwningExprResult ActOnStringLiteral(const Token *Toks,
+ virtual OwningExprResult ActOnStringLiteral(Scope *S, const Token *Toks,
unsigned NumToks) {
return ExprEmpty();
}
diff --git a/include/clang/Sema/Sema.h b/include/clang/Sema/Sema.h
index 5ef8e5f..cbe292f 100644
--- a/include/clang/Sema/Sema.h
+++ b/include/clang/Sema/Sema.h
@@ -1985,7 +1985,7 @@ public:
/// ActOnStringLiteral - The specified tokens were lexed as pasted string
/// fragments (e.g. "foo" "bar" L"baz").
- virtual OwningExprResult ActOnStringLiteral(const Token *Toks,
+ virtual OwningExprResult ActOnStringLiteral(Scope *S, const Token *Toks,
unsigned NumToks);
// Binary/Unary Operators. 'Tok' is the token for the operator.
@@ -2975,6 +2975,9 @@ public:
bool CheckLiteralOperatorDeclaration(FunctionDecl *FnDecl);
+ OwningExprResult BuildUDStringLiteral(Scope *S, StringLiteral *SL, unsigned L,
+ IdentifierInfo *II);
+
//===--------------------------------------------------------------------===//
// C++ Templates [C++ 14]
//
diff --git a/lib/AST/ExprConstant.cpp b/lib/AST/ExprConstant.cpp
index 3b288ec..195ed57 100644
--- a/lib/AST/ExprConstant.cpp
+++ b/lib/AST/ExprConstant.cpp
@@ -2419,7 +2419,8 @@ static ICEDiag CheckICE(const Expr* E, ASTContext &Ctx) {
case Expr::UnaryTypeTraitExprClass:
return NoDiag();
case Expr::CallExprClass:
- case Expr::CXXOperatorCallExprClass: {
+ case Expr::CXXOperatorCallExprClass:
+ case Expr::UDStringLiteralClass: {
const CallExpr *CE = cast<CallExpr>(E);
if (CE->isBuiltinCall(Ctx))
return CheckEvalInICE(E, Ctx);
diff --git a/lib/AST/StmtPrinter.cpp b/lib/AST/StmtPrinter.cpp
index 7f497a3..0f3ffa0 100644
--- a/lib/AST/StmtPrinter.cpp
+++ b/lib/AST/StmtPrinter.cpp
@@ -1197,6 +1197,11 @@ void StmtPrinter::VisitUnresolvedMemberExpr(UnresolvedMemberExpr *Node) {
}
}
+void StmtPrinter::VisitUDStringLiteral(UDStringLiteral *Node) {
+ VisitStringLiteral(Node->getBaseLiteral());
+ OS << Node->getUDSuffix()->getName();
+}
+
static const char *getTypeTraitName(UnaryTypeTrait UTT) {
switch (UTT) {
default: assert(false && "Unknown type trait");
diff --git a/lib/AST/StmtProfile.cpp b/lib/AST/StmtProfile.cpp
index 5beefd1..dd51376 100644
--- a/lib/AST/StmtProfile.cpp
+++ b/lib/AST/StmtProfile.cpp
@@ -827,6 +827,12 @@ void StmtProfiler::VisitUnresolvedMemberExpr(UnresolvedMemberExpr *S) {
VisitTemplateArguments(S->getTemplateArgs(), S->getNumTemplateArgs());
}
+void StmtProfiler::VisitUDStringLiteral(UDStringLiteral *S) {
+ VisitExpr(S);
+ VisitStringLiteral(S->getBaseLiteral());
+ ID.AddString(S->getUDSuffix()->getName());
+}
+
void StmtProfiler::VisitObjCStringLiteral(ObjCStringLiteral *S) {
VisitExpr(S);
}
diff --git a/lib/Lex/Lexer.cpp b/lib/Lex/Lexer.cpp
index 9a96934..079c327 100644
--- a/lib/Lex/Lexer.cpp
+++ b/lib/Lex/Lexer.cpp
@@ -547,6 +547,11 @@ static void InitCharacterInfo() {
isInited = true;
}
+/// isIdentifierStart - Return true if this is the start character of an
+/// identifier, which is [a-zA-Z_].
+static inline bool isIdentifierStart(unsigned char c) {
+ return (CharInfo[c] & (CHAR_LETTER|CHAR_UNDER)) ? true : false;
+}
/// isIdentifierBody - Return true if this is the body character of an
/// identifier, which is [a-zA-Z0-9_].
@@ -980,8 +985,30 @@ void Lexer::LexStringLiteral(Token &Result, const char *CurPtr, bool Wide) {
// Update the location of the token as well as the BufferPtr instance var.
const char *TokStart = BufferPtr;
- FormTokenWithChars(Result, CurPtr,
- Wide ? tok::wide_string_literal : tok::string_literal);
+ tok::TokenKind Kind = Wide ? tok::wide_string_literal : tok::string_literal;
+
+ // FIXME: Handle UCNs
+ unsigned Size;
+ if (PP && PP->getLangOptions().CPlusPlus0x &&
+ isIdentifierStart(getCharAndSize(CurPtr, Size))) {
+ Result.makeUserDefinedLiteral(ExtraDataAllocator);
+ Result.setFlagValue(Token::SLPortionClean, !Result.needsCleaning());
+ Result.setKind(Kind);
+ Result.setLiteralLength(CurPtr - BufferPtr);
+
+ // FIXME: We hack around the lexer's routines a lot here.
+ BufferPtr = CurPtr;
+ bool OldRawMode = LexingRawMode;
+ LexingRawMode = true;
+ LexIdentifier(Result, ConsumeChar(CurPtr, Size, Result));
+ LexingRawMode = OldRawMode;
+ PP->LookUpIdentifierInfo(Result, CurPtr);
+
+ CurPtr = BufferPtr;
+ BufferPtr = TokStart;
+ }
+
+ FormTokenWithChars(Result, CurPtr, Kind);
Result.setLiteralData(TokStart);
}
diff --git a/lib/Lex/LiteralSupport.cpp b/lib/Lex/LiteralSupport.cpp
index a12c4ae..eb7337a 100644
--- a/lib/Lex/LiteralSupport.cpp
+++ b/lib/Lex/LiteralSupport.cpp
@@ -758,30 +758,38 @@ CharLiteralParser::CharLiteralParser(const char *begin, const char *end,
///
StringLiteralParser::
StringLiteralParser(const Token *StringToks, unsigned NumStringToks,
- Preprocessor &pp, bool Complain) : PP(pp) {
+ Preprocessor &pp, bool Complain) : PP(pp), hadError(false) {
// Scan all of the string portions, remember the max individual token length,
// computing a bound on the concatenated string length, and see whether any
// piece is a wide-string. If any of the string portions is a wide-string
// literal, the result is a wide-string literal [C99 6.4.5p4].
- MaxTokenLength = StringToks[0].getLength();
- SizeBound = StringToks[0].getLength()-2; // -2 for "".
+ MaxTokenLength = StringToks[0].getLiteralLength();
+ SizeBound = StringToks[0].getLiteralLength()-2; // -2 for "".
AnyWide = StringToks[0].is(tok::wide_string_literal);
-
- hadError = false;
+ UDSuffix = StringToks[0].getIdentifierInfo();
// Implement Translation Phase #6: concatenation of string literals
/// (C99 5.1.1.2p1). The common case is only one string fragment.
for (unsigned i = 1; i != NumStringToks; ++i) {
// The string could be shorter than this if it needs cleaning, but this is a
// reasonable bound, which is all we need.
- SizeBound += StringToks[i].getLength()-2; // -2 for "".
+ SizeBound += StringToks[i].getLiteralLength()-2; // -2 for "".
// Remember maximum string piece length.
- if (StringToks[i].getLength() > MaxTokenLength)
- MaxTokenLength = StringToks[i].getLength();
+ if (StringToks[i].getLiteralLength() > MaxTokenLength)
+ MaxTokenLength = StringToks[i].getLiteralLength();
// Remember if we see any wide strings.
AnyWide |= StringToks[i].is(tok::wide_string_literal);
+
+ if (StringToks[i].isUserDefinedLiteral()) {
+ if (UDSuffix && UDSuffix != StringToks[i].getIdentifierInfo()) {
+ // FIXME: Improve location and note previous
+ PP.Diag(StringToks[0].getLocation(), diag::err_ud_suffix_mismatch);
+ hadError = true;
+ } else if (!UDSuffix)
+ UDSuffix = StringToks[0].getIdentifierInfo();
+ }
}
// Include space for the null terminator.
@@ -823,7 +831,7 @@ StringLiteralParser(const Token *StringToks, unsigned NumStringToks,
// and 'spelled' tokens can only shrink.
bool StringInvalid = false;
unsigned ThisTokLen = PP.getSpelling(StringToks[i], ThisTokBuf,
- &StringInvalid);
+ &StringInvalid, true);
if (StringInvalid) {
hadError = 1;
continue;
@@ -938,7 +946,7 @@ unsigned StringLiteralParser::getOffsetOfStringByte(const Token &Tok,
bool Complain) {
// Get the spelling of the token.
llvm::SmallString<16> SpellingBuffer;
- SpellingBuffer.resize(Tok.getLength());
+ SpellingBuffer.resize(Tok.getLiteralLength());
bool StringInvalid = false;
const char *SpellingPtr = &SpellingBuffer[0];
diff --git a/lib/Lex/Preprocessor.cpp b/lib/Lex/Preprocessor.cpp
index 77b88f7..df4a22c 100644
--- a/lib/Lex/Preprocessor.cpp
+++ b/lib/Lex/Preprocessor.cpp
@@ -344,15 +344,25 @@ std::string Preprocessor::getSpelling(const Token &Tok, bool *Invalid) const {
/// to point to a constant buffer with the data already in it (avoiding a
/// copy). The caller is not allowed to modify the returned buffer pointer
/// if an internal buffer is returned.
-unsigned Preprocessor::getSpelling(const Token &Tok,
- const char *&Buffer, bool *Invalid) const {
+///
+/// If LiteralOnly is specified, only the literal portion of the token is
+/// processed.
+unsigned Preprocessor::getSpelling(const Token &Tok, const char *&Buffer,
+ bool *Invalid, bool LiteralOnly) const {
assert((int)Tok.getLength() >= 0 && "Token character range is bogus!");
+ assert((!LiteralOnly || Tok.isLiteral()) &&
+ "LiteralOnly used on a non-literal token");
+
+ unsigned (Token::*getLength) () const = LiteralOnly
+ ? &Token::getLiteralLength : &Token::getLength;
// If this token is an identifier, just return the string from the identifier
// table, which is very quick.
if (const IdentifierInfo *II = Tok.getIdentifierInfo()) {
- Buffer = II->getNameStart();
- return II->getLength();
+ if (!Tok.isUserDefinedLiteral()) {
+ Buffer = II->getNameStart();
+ return II->getLength();
+ }
}
// Otherwise, compute the start of the token in the input lexer buffer.
@@ -373,20 +383,20 @@ unsigned Preprocessor::getSpelling(const Token &Tok,
}
// If this token contains nothing interesting, return it directly.
- if (!Tok.needsCleaning()) {
+ if (!(LiteralOnly ? Tok.literalNeedsCleaning() : Tok.needsCleaning())) {
Buffer = TokStart;
- return Tok.getLength();
+ return (Tok.*getLength)();
}
// Otherwise, hard case, relex the characters into the string.
char *OutBuf = const_cast<char*>(Buffer);
- for (const char *Ptr = TokStart, *End = TokStart+Tok.getLength();
+ for (const char *Ptr = TokStart, *End = TokStart+(Tok.*getLength)();
Ptr != End; ) {
unsigned CharSize;
*OutBuf++ = Lexer::getCharAndSizeNoWarn(Ptr, CharSize, Features);
Ptr += CharSize;
}
- assert(unsigned(OutBuf-Buffer) != Tok.getLength() &&
+ assert(unsigned(OutBuf-Buffer) != (Tok.*getLength)() &&
"NeedsCleaning flag set on something that didn't need cleaning!");
return OutBuf-Buffer;
diff --git a/lib/Parse/ParseExpr.cpp b/lib/Parse/ParseExpr.cpp
index 844dd3f..0cc6877 100644
--- a/lib/Parse/ParseExpr.cpp
+++ b/lib/Parse/ParseExpr.cpp
@@ -1572,7 +1572,8 @@ Parser::OwningExprResult Parser::ParseStringLiteralExpression() {
} while (isTokenStringLiteral());
// Pass the set of string tokens, ready for concatenation, to the actions.
- return Actions.ActOnStringLiteral(&StringToks[0], StringToks.size());
+ return Actions.ActOnStringLiteral(getCurScope(), &StringToks[0],
+ StringToks.size());
}
/// ParseExpressionList - Used for C/C++ (argument-)expression-list.
diff --git a/lib/Parse/ParseStmt.cpp b/lib/Parse/ParseStmt.cpp
index a83aff1..5038da5 100644
--- a/lib/Parse/ParseStmt.cpp
+++ b/lib/Parse/ParseStmt.cpp
@@ -1233,11 +1233,11 @@ Parser::OwningStmtResult Parser::FuzzyParseMicrosoftAsmStatement() {
Tok.isNot(tok::eof));
}
Token t;
+ t.startToken();
t.setKind(tok::string_literal);
t.setLiteralData("\"FIXME: not done\"");
- t.clearFlag(Token::NeedsCleaning);
t.setLength(17);
- OwningExprResult AsmString(Actions.ActOnStringLiteral(&t, 1));
+ OwningExprResult AsmString(Actions.ActOnStringLiteral(getCurScope(), &t, 1));
ExprVector Constraints(Actions);
ExprVector Exprs(Actions);
ExprVector Clobbers(Actions);
diff --git a/lib/Sema/SemaDeclCXX.cpp b/lib/Sema/SemaDeclCXX.cpp
index a1a8466..1aba0ef 100644
--- a/lib/Sema/SemaDeclCXX.cpp
+++ b/lib/Sema/SemaDeclCXX.cpp
@@ -5805,6 +5805,9 @@ bool Sema::CheckLiteralOperatorDeclaration(FunctionDecl *FnDecl) {
return true;
}
+ if (FnDecl->getDeclName().getCXXLiteralIdentifier()->getName()[0] != '_')
+ Diag(FnDecl->getLocation(), diag::warn_literal_operator_no_underscore);
+
bool Valid = false;
// template <char...> type operator "" name() is the only valid template
diff --git a/lib/Sema/SemaExpr.cpp b/lib/Sema/SemaExpr.cpp
index 59d0328..b581272 100644
--- a/lib/Sema/SemaExpr.cpp
+++ b/lib/Sema/SemaExpr.cpp
@@ -372,7 +372,8 @@ QualType Sema::UsualArithmeticConversions(Expr *&lhsExpr, Expr *&rhsExpr,
/// string.
///
Action::OwningExprResult
-Sema::ActOnStringLiteral(const Token *StringToks, unsigned NumStringToks) {
+Sema::ActOnStringLiteral(Scope *S, const Token *StringToks,
+ unsigned NumStringToks) {
assert(NumStringToks && "Must have at least one string!");
StringLiteralParser Literal(StringToks, NumStringToks, PP);
@@ -398,12 +399,18 @@ Sema::ActOnStringLiteral(const Token *StringToks, unsigned NumStringToks) {
llvm::APInt(32, Literal.GetNumStringChars()+1),
ArrayType::Normal, 0);
+ StringLiteral *SL = StringLiteral::Create(Context, Literal.GetString(),
+ Literal.GetStringLength(),
+ Literal.AnyWide, StrTy,
+ &StringTokLocs[0],
+ StringTokLocs.size());
+
+ if (Literal.isUserDefinedLiteral())
+ return BuildUDStringLiteral(S, SL, Literal.GetNumStringChars(),
+ Literal.getUDSuffix());
+
// Pass &StringTokLocs[0], StringTokLocs.size() to factory!
- return Owned(StringLiteral::Create(Context, Literal.GetString(),
- Literal.GetStringLength(),
- Literal.AnyWide, StrTy,
- &StringTokLocs[0],
- StringTokLocs.size()));
+ return Owned(SL);
}
/// ShouldSnapshotBlockValueReference - Return true if a reference inside of
diff --git a/lib/Sema/SemaExprCXX.cpp b/lib/Sema/SemaExprCXX.cpp
index 35e679f..153eef3 100644
--- a/lib/Sema/SemaExprCXX.cpp
+++ b/lib/Sema/SemaExprCXX.cpp
@@ -3060,3 +3060,55 @@ Sema::OwningExprResult Sema::ActOnFinishFullExpr(ExprArg Arg) {
return Owned(FullExpr);
}
+
+Sema::OwningExprResult Sema::BuildUDStringLiteral(Scope *S, StringLiteral *SL,
+ unsigned L,
+ IdentifierInfo *II) {
+ DeclarationName DN = Context.DeclarationNames.getCXXLiteralOperatorName(II);
+
+ LookupResult R(*this, DN, SL->getLocStart(), LookupOrdinaryName);
+ LookupName(R, S);
+
+ llvm::APInt APL(Context.getTypeSize(Context.getSizeType()), L);
+
+ Expr *Args[2];
+ Args[0] = SL;
+ Args[1] = new (Context) IntegerLiteral(APL, Context.getSizeType(),
+ SourceLocation());
+
+ OverloadCandidateSet CandidateSet(SL->getLocStart());
+ AddFunctionCandidates(R.asUnresolvedSet(), Args, 2, CandidateSet);
+ OverloadCandidateSet::iterator Best;
+ OverloadingResult Result = BestViableFunction(CandidateSet,
+ SL->getLocStart(), Best);
+
+ if (Result != OR_Success) {
+ Diag(SL->getLocStart(), diag::err_literal_operator_overload)
+ << SL->getSourceRange() << II->getName();
+ return ExprError();
+ }
+
+ assert(Best->Function && "Literal operator function not a real function");
+ FunctionDecl *FD = Best->Function;
+
+ OwningExprResult InputInit
+ = PerformCopyInitialization(InitializedEntity::InitializeParameter(
+ FD->getParamDecl(0)),
+ SourceLocation(), Owned(SL));
+ if (InputInit.isInvalid())
+ return ExprError();
+ Args[0] = InputInit.takeAs<Expr>();
+
+ QualType ResultTy = FD->getResultType().getNonReferenceType();
+ Expr *Fn = new (Context) DeclRefExpr(FD, FD->getType(), SourceLocation());
+ UsualUnaryConversions(Fn);
+
+ ExprOwningPtr<CallExpr> TheCall(this,
+ new (Context) UDStringLiteral(Context, SL, Fn, Args, 2, ResultTy));
+
+ if (CheckCallReturnType(FD->getResultType(), SL->getLocStart(),
+ TheCall.get(), FD))
+ return ExprError();
+
+ return MaybeBindToTemporary(TheCall.release());
+}
diff --git a/lib/Sema/TreeTransform.h b/lib/Sema/TreeTransform.h
index 2f8d075..0f12b5c 100644
--- a/lib/Sema/TreeTransform.h
+++ b/lib/Sema/TreeTransform.h
@@ -6043,6 +6043,12 @@ TreeTransform<Derived>::TransformObjCEncodeExpr(ObjCEncodeExpr *E) {
template<typename Derived>
Sema::OwningExprResult
+TreeTransform<Derived>::TransformUDStringLiteral(UDStringLiteral *E) {
+ return SemaRef.Owned(E->Retain());
+}
+
+template<typename Derived>
+Sema::OwningExprResult
TreeTransform<Derived>::TransformObjCMessageExpr(ObjCMessageExpr *E) {
// Transform arguments.
bool ArgChanged = false;
diff --git a/test/Parser/cxx0x-literal-operators.cpp b/test/Parser/cxx0x-literal-operators.cpp
index 30b2903..cc979b3 100644
--- a/test/Parser/cxx0x-literal-operators.cpp
+++ b/test/Parser/cxx0x-literal-operators.cpp
@@ -1,5 +1,5 @@
// RUN: %clang_cc1 -fsyntax-only -verify -std=c++0x %s
void operator "" (const char *); // expected-error {{expected identifier}}
-void operator "k" foo(const char *); // expected-error {{string literal after 'operator' must be '""'}}
-void operator "" tester (const char *);
+void operator "k" _foo(const char *); // expected-error {{string literal after 'operator' must be '""'}}
+void operator "" _tester (const char *);
diff --git a/test/SemaCXX/literal-operator-dcls.cpp b/test/SemaCXX/literal-operator-dcls.cpp
new file mode 100644
index 0000000..88bf2b9
--- /dev/null
+++ b/test/SemaCXX/literal-operator-dcls.cpp
@@ -0,0 +1,46 @@
+// RUN: %clang_cc1 -fsyntax-only -verify -std=c++0x %s
+
+#include <stddef.h>
+
+struct tag {
+ void operator "" _tag_bad (const char *); // expected-error {{literal operator 'operator "" _tag_bad' must be in a namespace or global scope}}
+ friend void operator "" _tag_good (const char *);
+};
+
+namespace ns { void operator "" _ns_good (const char *); }
+
+// Check extern "C++" declarations
+extern "C++" void operator "" _extern_good (const char *);
+extern "C++" { void operator "" _extern_good (const char *); }
+
+void fn () { void operator "" _fn_bad (const char *); } // expected-error {{literal operator 'operator "" _fn_bad' must be in a namespace or global scope}}
+
+// Warning name
+void operator "" warn (const char *); // expected-warning {{reserved for future standardization}}
+
+// One-param declarations (const char * was already checked)
+void operator "" _good (char);
+void operator "" _good (wchar_t);
+void operator "" _good (char16_t);
+void operator "" _good (char32_t);
+void operator "" _good (unsigned long long);
+void operator "" _good (long double);
+
+// Two-param declarations
+void operator "" _good (const char *, size_t);
+void operator "" _good (const wchar_t *, size_t);
+void operator "" _good (const char16_t *, size_t);
+void operator "" _good (const char32_t *, size_t);
+
+// Check typedef and array equivalences
+void operator "" _good (const char[]);
+typedef const char c;
+void operator "" _good (c*);
+
+// Check extra cv-qualifiers
+void operator "" _cv_good (volatile const char *, const size_t);
+
+// Template delcaration (not implemented yet)
+// template <char...> void operator "" good ();
+
+// FIXME: Test some invalid decls that might crop up.
diff --git a/test/SemaCXX/literal-operators.cpp b/test/SemaCXX/literal-operators.cpp
index ec585a6..93fd4b6 100644
--- a/test/SemaCXX/literal-operators.cpp
+++ b/test/SemaCXX/literal-operators.cpp
@@ -2,42 +2,21 @@
#include <stddef.h>
-struct tag {
- void operator "" tag_bad (const char *); // expected-error {{literal operator 'operator "" tag_bad' must be in a namespace or global scope}}
- friend void operator "" tag_good (const char *);
+template <typename T, typename U> struct same_type {
+ static const bool value = false;
};
-namespace ns { void operator "" ns_good (const char *); }
-
-// Check extern "C++" declarations
-extern "C++" void operator "" extern_good (const char *);
-extern "C++" { void operator "" extern_good (const char *); }
-
-void fn () { void operator "" fn_bad (const char *); } // expected-error {{literal operator 'operator "" fn_bad' must be in a namespace or global scope}}
-
-// One-param declarations (const char * was already checked)
-void operator "" good (char);
-void operator "" good (wchar_t);
-void operator "" good (char16_t);
-void operator "" good (char32_t);
-void operator "" good (unsigned long long);
-void operator "" good (long double);
+template <typename T> struct same_type<T, T> {
+ static const bool value = true;
+};
-// Two-param declarations
-void operator "" good (const char *, size_t);
-void operator "" good (const wchar_t *, size_t);
-void operator "" good (const char16_t *, size_t);
-void operator "" good (const char32_t *, size_t);
+int operator "" _int (const char *, size_t);
+static_assert(same_type<int, decltype(""_int)>::value, "not the same type!");
-// Check typedef and array equivalences
-void operator "" good (const char[]);
-typedef const char c;
-void operator "" good (c*);
+int i = ""_int;
+int j = L""_int; // expected-error {{no matching literal operator function}}
-// Check extra cv-qualifiers
-void operator "" cv_good (volatile const char *, const size_t);
+int operator "" _int (const wchar_t *, size_t);
-// Template delcaration (not implemented yet)
-// template <char...> void operator "" good ();
+int k = L""_int;
-// FIXME: Test some invalid decls that might crop up.
diff --git a/tools/libclang/CXCursor.cpp b/tools/libclang/CXCursor.cpp
index be3623f..04d2b78 100644
--- a/tools/libclang/CXCursor.cpp
+++ b/tools/libclang/CXCursor.cpp
@@ -218,6 +218,7 @@ CXCursor cxcursor::MakeCXCursor(Stmt *S, Decl *Parent, ASTUnit *TU) {
case Stmt::CXXMemberCallExprClass:
case Stmt::CXXConstructExprClass:
case Stmt::CXXTemporaryObjectExprClass:
+ case Stmt::UDStringLiteralClass:
// FIXME: CXXUnresolvedConstructExpr
// FIXME: ObjCImplicitSetterGetterRefExpr?
K = CXCursor_CallExpr;
_______________________________________________
cfe-commits mailing list
[email protected]
http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
