Fix whitespace handling in ## operator

In x ## y, where x and y are regular tokens, whitespace between x and ##
is ignored, and whitespace between ## and y is also ignored. When either
x or y is a function argument, whitespace was preserved, but it should
not be. This patch removes the checks for whitespace before ## and
before y, and in the special case where x is an empty macro argument and
y is a regular token, actively removes whitespace before y.

One existing test is affected by that change, but as clang's output now
matches the standard's requirements and that of GCC, I've tweaked the
testcase.

Patch by Harald van Dijk!

git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@200785 91177308-0d34-0410-b5e6-96231b3b80d8

lib/Lex/TokenLexer.cpp

@@ -223,6 +223,13 @@ void TokenLexer::ExpandFunctionArguments() {
       continue;
     }
 
+    // Find out if there is a paste (##) operator before or after the token.
+    bool NonEmptyPasteBefore =
+      !ResultToks.empty() && ResultToks.back().is(tok::hashhash);
+    bool PasteBefore = i != 0 && Tokens[i-1].is(tok::hashhash);
+    bool PasteAfter = i+1 != e && Tokens[i+1].is(tok::hashhash);
+    assert(!NonEmptyPasteBefore || PasteBefore);
+
     // Otherwise, if this is not an argument token, just add the token to the
     // output buffer.
     IdentifierInfo *II = CurTok.getIdentifierInfo();
@@ -234,7 +241,9 @@ void TokenLexer::ExpandFunctionArguments() {
       if (NextTokGetsSpace) {
         ResultToks.back().setFlag(Token::LeadingSpace);
         NextTokGetsSpace = false;
-      }
+      } else if (PasteBefore && !NonEmptyPasteBefore)
+        ResultToks.back().clearFlag(Token::LeadingSpace);
+
       continue;
     }
 
@@ -242,13 +251,7 @@ void TokenLexer::ExpandFunctionArguments() {
     // input.
     MadeChange = true;
 
-    // Otherwise, this is a use of the argument.  Find out if there is a paste
-    // (##) operator before or after the argument.
-    bool NonEmptyPasteBefore =
-      !ResultToks.empty() && ResultToks.back().is(tok::hashhash);
-    bool PasteBefore = i != 0 && Tokens[i-1].is(tok::hashhash);
-    bool PasteAfter = i+1 != e && Tokens[i+1].is(tok::hashhash);
-    assert(!NonEmptyPasteBefore || PasteBefore);
+    // Otherwise, this is a use of the argument.
 
     // In Microsoft mode, remove the comma before __VA_ARGS__ to ensure there
     // are no trailing commas if __VA_ARGS__ is empty.
@@ -358,8 +361,8 @@ void TokenLexer::ExpandFunctionArguments() {
       // assembler-with-cpp mode, invalid pastes are allowed through: in this
       // case, we do not want the extra whitespace to be added.  For example,
       // we want ". ## foo" -> ".foo" not ". foo".
-      if ((CurTok.hasLeadingSpace() || NextTokGetsSpace) &&
-          !NonEmptyPasteBefore)
+      if ((CurTok.hasLeadingSpace() && !PasteBefore) ||
+          (NextTokGetsSpace && !NonEmptyPasteBefore))
         ResultToks[ResultToks.size()-NumToks].setFlag(Token::LeadingSpace);
 
       NextTokGetsSpace = false;
@@ -370,11 +373,11 @@ void TokenLexer::ExpandFunctionArguments() {
     // 6.10.3.3p2,3) calls for a bunch of placemarker stuff to occur.  We
     // implement this by eating ## operators when a LHS or RHS expands to
     // empty.
-    NextTokGetsSpace |= CurTok.hasLeadingSpace();
+    if (!PasteBefore)
+      NextTokGetsSpace |= i != 0 && CurTok.hasLeadingSpace();
     if (PasteAfter) {
       // Discard the argument token and skip (don't copy to the expansion
       // buffer) the paste operator after it.
-      NextTokGetsSpace |= Tokens[i+1].hasLeadingSpace();
       ++i;
       continue;
     }
@@ -385,7 +388,7 @@ void TokenLexer::ExpandFunctionArguments() {
     assert(PasteBefore);
     if (NonEmptyPasteBefore) {
       assert(ResultToks.back().is(tok::hashhash));
-      NextTokGetsSpace |= ResultToks.pop_back_val().hasLeadingSpace();
+      ResultToks.pop_back();
     }
 
     // If this is the __VA_ARGS__ token, and if the argument wasn't provided,

test/Preprocessor/macro_paste_commaext.c

@@ -1,8 +1,8 @@
 // RUN: %clang_cc1 %s -E | grep 'V);'
 // RUN: %clang_cc1 %s -E | grep 'W, 1, 2);'
 // RUN: %clang_cc1 %s -E | grep 'X, 1, 2);'
-// RUN: %clang_cc1 %s -E | grep 'Y, );'
-// RUN: %clang_cc1 %s -E | grep 'Z, );'
+// RUN: %clang_cc1 %s -E | grep 'Y,);'
+// RUN: %clang_cc1 %s -E | grep 'Z,);'
 
 #define debug(format, ...) format, ## __VA_ARGS__)
 debug(V);

test/Preprocessor/macro_paste_spacing.c

@@ -1,7 +1,21 @@
-// RUN: %clang_cc1 %s -E | grep "^xy$"
+// RUN: %clang_cc1 -E %s | FileCheck --strict-whitespace %s
 
 #define A  x ## y
 blah
 
 A
+// CHECK: {{^}}xy{{$}}
 
+#define B(x, y) [v ## w] [ v##w] [v##w ] [w ## x] [ w##x] [w##x ] [x ## y] [ x##y] [x##y ] [y ## z] [ y##z] [y##z ]
+B(x,y)
+// CHECK: [vw] [ vw] [vw ] [wx] [ wx] [wx ] [xy] [ xy] [xy ] [yz] [ yz] [yz ]
+B(x,)
+// CHECK: [vw] [ vw] [vw ] [wx] [ wx] [wx ] [x] [ x] [x ] [z] [ z] [z ]
+B(,y)
+// CHECK: [vw] [ vw] [vw ] [w] [ w] [w ] [y] [ y] [y ] [yz] [ yz] [yz ]
+B(,)
+// CHECK: [vw] [ vw] [vw ] [w] [ w] [w ] [] [ ] [ ] [z] [ z] [z ]
+
+#define C(x, y, z) [x ## y ## z]
+C(,,) C(a,,) C(,b,) C(,,c) C(a,b,) C(a,,c) C(,b,c) C(a,b,c)
+// CHECK: [] [a] [b] [c] [ab] [ac] [bc] [abc]