Small cleanup of literal semantic analysis: hiding 'char *' pointers behind

StringRef makes code cleaner.  Also, make the temporary buffer smaller:
512 characters is unreasonably large for integer literals.

git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@164484 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/include/clang/Lex/LiteralSupport.h b/include/clang/Lex/LiteralSupport.h
index fec158d7a8983291cec6d3954eff96f652a177d0..3b68d1b570f4efc83de54544288a068e76a09310 100644 (file)
--- a/include/clang/Lex/LiteralSupport.h
+++ b/include/clang/Lex/LiteralSupport.h
@@ -18,6 +18,7 @@
 #include "clang/Basic/LLVM.h"
 #include "llvm/ADT/APFloat.h"
 #include "llvm/ADT/SmallString.h"
+#include "llvm/ADT/StringRef.h"
 #include "llvm/Support/DataTypes.h"
 #include "clang/Basic/TokenKinds.h"
 #include <cctype>
@@ -48,8 +49,9 @@ class NumericLiteralParser {
   bool saw_exponent, saw_period, saw_ud_suffix;
 
 public:
-  NumericLiteralParser(const char *begin, const char *end,
-                       SourceLocation Loc, Preprocessor &PP);
+  NumericLiteralParser(StringRef TokSpelling,
+                       SourceLocation TokLoc,
+                       Preprocessor &PP);
   bool hadError;
   bool isUnsigned;
   bool isLong;        // This is *not* set for long long.

diff --git a/lib/Lex/LiteralSupport.cpp b/lib/Lex/LiteralSupport.cpp
index be53429e6f2d8da5f37875dcf02d74c00aaaa199..e65d2edbd32ec0aa8652d8afde7b6fd752d817df 100644 (file)
--- a/lib/Lex/LiteralSupport.cpp
+++ b/lib/Lex/LiteralSupport.cpp
@@ -447,19 +447,19 @@ static void EncodeUCNEscape(const char *ThisTokBegin, const char *&ThisTokBuf,
 ///       floating-constant: [C99 6.4.4.2]
 ///         TODO: add rules...
 ///
-NumericLiteralParser::
-NumericLiteralParser(const char *begin, const char *end,
-                     SourceLocation TokLoc, Preprocessor &pp)
-  : PP(pp), ThisTokBegin(begin), ThisTokEnd(end) {
+NumericLiteralParser::NumericLiteralParser(StringRef TokSpelling,
+                                           SourceLocation TokLoc,
+                                           Preprocessor &PP)
+  : PP(PP), ThisTokBegin(TokSpelling.begin()), ThisTokEnd(TokSpelling.end()) {
 
   // This routine assumes that the range begin/end matches the regex for integer
   // and FP constants (specifically, the 'pp-number' regex), and assumes that
   // the byte at "*end" is both valid and not part of the regex.  Because of
   // this, it doesn't have to check for 'overscan' in various places.
-  assert(!isalnum(*end) && *end != '.' && *end != '_' &&
+  assert(!isalnum(*ThisTokEnd) && *ThisTokEnd != '.' && *ThisTokEnd != '_' &&
          "Lexer didn't maximally munch?");
 
-  s = DigitsBegin = begin;
+  s = DigitsBegin = ThisTokBegin;
   saw_exponent = false;
   saw_period = false;
   saw_ud_suffix = false;
@@ -481,7 +481,7 @@ NumericLiteralParser(const char *begin, const char *end,
     if (s == ThisTokEnd) {
       // Done.
     } else if (isxdigit(*s) && !(*s == 'e' || *s == 'E')) {
-      PP.Diag(PP.AdvanceToTokenCharacter(TokLoc, s-begin),
+      PP.Diag(PP.AdvanceToTokenCharacter(TokLoc, s - ThisTokBegin),
               diag::err_invalid_decimal_digit) << StringRef(s, 1);
       hadError = true;
       return;
@@ -499,7 +499,7 @@ NumericLiteralParser(const char *begin, const char *end,
       if (first_non_digit != s) {
         s = first_non_digit;
       } else {
-        PP.Diag(PP.AdvanceToTokenCharacter(TokLoc, Exponent-begin),
+        PP.Diag(PP.AdvanceToTokenCharacter(TokLoc, Exponent - ThisTokBegin),
                 diag::err_exponent_has_no_digits);
         hadError = true;
         return;
@@ -595,7 +595,7 @@ NumericLiteralParser(const char *begin, const char *end,
     case 'j':
     case 'J':
       if (isImaginary) break;   // Cannot be repeated.
-      PP.Diag(PP.AdvanceToTokenCharacter(TokLoc, s-begin),
+      PP.Diag(PP.AdvanceToTokenCharacter(TokLoc, s - ThisTokBegin),
               diag::ext_imaginary_constant);
       isImaginary = true;
       continue;  // Success.
@@ -613,7 +613,7 @@ NumericLiteralParser(const char *begin, const char *end,
     }
 
     // Report an error if there are any.
-    PP.Diag(PP.AdvanceToTokenCharacter(TokLoc, SuffixBegin-begin),
+    PP.Diag(PP.AdvanceToTokenCharacter(TokLoc, SuffixBegin - ThisTokBegin),
             isFPConstant ? diag::err_invalid_suffix_float_constant :
                            diag::err_invalid_suffix_integer_constant)
       << StringRef(SuffixBegin, ThisTokEnd-SuffixBegin);
@@ -649,7 +649,7 @@ void NumericLiteralParser::ParseNumberStartingWithZero(SourceLocation TokLoc) {
     }
 
     if (noSignificand) {
-      PP.Diag(PP.AdvanceToTokenCharacter(TokLoc, s-ThisTokBegin), \
+      PP.Diag(PP.AdvanceToTokenCharacter(TokLoc, s - ThisTokBegin),
         diag::err_hexconstant_requires_digits);
       hadError = true;
       return;

diff --git a/lib/Lex/PPExpressions.cpp b/lib/Lex/PPExpressions.cpp
index 49dfa198f638bd80ec3ab32ef4db7c7bf3529f76..f5984f58bf6f33cc1882ce3e30626b8101c95272 100644 (file)
--- a/lib/Lex/PPExpressions.cpp
+++ b/lib/Lex/PPExpressions.cpp
@@ -206,8 +206,7 @@ static bool EvaluateValue(PPValue &Result, Token &PeekTok, DefinedTracker &DT,
     if (NumberInvalid)
       return true; // a diagnostic was already reported
 
-    NumericLiteralParser Literal(Spelling.begin(), Spelling.end(),
-                                 PeekTok.getLocation(), PP);
+    NumericLiteralParser Literal(Spelling, PeekTok.getLocation(), PP);
     if (Literal.hadError)
       return true; // a diagnostic was already reported.
 

diff --git a/lib/Sema/SemaExpr.cpp b/lib/Sema/SemaExpr.cpp
index ff7e9101f1c67cbdf467af2050d1c26e427f8782..b864a2e8a06e07195e891d428e0eb1dbea15b4b8 100644 (file)
--- a/lib/Sema/SemaExpr.cpp
+++ b/lib/Sema/SemaExpr.cpp
@@ -2644,19 +2644,20 @@ ExprResult Sema::ActOnNumericConstant(const Token &Tok, Scope *UDLScope) {
     return ActOnIntegerConstant(Tok.getLocation(), Val-'0');
   }
 
-  SmallString<512> IntegerBuffer;
-  // Add padding so that NumericLiteralParser can overread by one character.
-  IntegerBuffer.resize(Tok.getLength()+1);
-  const char *ThisTokBegin = &IntegerBuffer[0];
+  SmallString<128> SpellingBuffer;
+  // NumericLiteralParser wants to overread by one character.  Add padding to
+  // the buffer in case the token is copied to the buffer.  If getSpelling()
+  // returns a StringRef to the memory buffer, it should have a null char at
+  // the EOF, so it is also safe.
+  SpellingBuffer.resize(Tok.getLength() + 1);
 
   // Get the spelling of the token, which eliminates trigraphs, etc.
   bool Invalid = false;
-  unsigned ActualLength = PP.getSpelling(Tok, ThisTokBegin, &Invalid);
+  StringRef TokSpelling = PP.getSpelling(Tok, SpellingBuffer, &Invalid);
   if (Invalid)
     return ExprError();
 
-  NumericLiteralParser Literal(ThisTokBegin, ThisTokBegin+ActualLength,
-                               Tok.getLocation(), PP);
+  NumericLiteralParser Literal(TokSpelling, Tok.getLocation(), PP);
   if (Literal.hadError)
     return ExprError();
 
@@ -2722,7 +2723,7 @@ ExprResult Sema::ActOnNumericConstant(const Token &Tok, Scope *UDLScope) {
           Context.CharTy, llvm::APInt(32, Length + 1),
           ArrayType::Normal, 0);
       Expr *Lit = StringLiteral::Create(
-          Context, StringRef(ThisTokBegin, Length), StringLiteral::Ascii,
+          Context, StringRef(TokSpelling.data(), Length), StringLiteral::Ascii,
           /*Pascal*/false, StrTy, &TokLoc, 1);
       return BuildLiteralOperatorCall(R, OpNameInfo,
                                       llvm::makeArrayRef(&Lit, 1), TokLoc);
@@ -2738,7 +2739,7 @@ ExprResult Sema::ActOnNumericConstant(const Token &Tok, Scope *UDLScope) {
       bool CharIsUnsigned = Context.CharTy->isUnsignedIntegerType();
       llvm::APSInt Value(CharBits, CharIsUnsigned);
       for (unsigned I = 0, N = Literal.getUDSuffixOffset(); I != N; ++I) {
-        Value = ThisTokBegin[I];
+        Value = TokSpelling[I];
         TemplateArgument Arg(Context, Value, Context.CharTy);
         TemplateArgumentLocInfo ArgInfo;
         ExplicitArgs.addArgument(TemplateArgumentLoc(Arg, ArgInfo));