[PATCH] Don't break line comments with escaped newlines.
Alexander Kornienko
alexfh at google.com
Tue Jul 16 13:52:48 PDT 2013
Removed extra penalty addition. Added a FIXME to the test.
Hi djasper,
http://llvm-reviews.chandlerc.com/D1146
CHANGE SINCE LAST DIFF
http://llvm-reviews.chandlerc.com/D1146?vs=2847&id=2848#toc
Files:
lib/Format/Format.cpp
lib/Format/TokenAnnotator.cpp
unittests/Format/FormatTest.cpp
Index: lib/Format/Format.cpp
===================================================================
--- lib/Format/Format.cpp
+++ lib/Format/Format.cpp
@@ -904,6 +904,11 @@
// Only break up default narrow strings.
if (!Current.TokenText.startswith("\""))
return 0;
+ // Don't break string literals with escaped newlines. As clang-format must
+ // not change the string's content, it is unlikely that we'll end up with
+ // a better format.
+ if (Current.TokenText.find("\\\n") != StringRef::npos)
+ return 0;
Token.reset(new BreakableStringLiteral(Current, StartColumn,
Line.InPPDirective, Encoding));
@@ -914,6 +919,23 @@
} else if (Current.Type == TT_LineComment &&
(Current.Previous == NULL ||
Current.Previous->Type != TT_ImplicitStringLiteral)) {
+ // Don't break line comments with escaped newlines. These look like
+ // separate line comments, but in fact contain a single line comment with
+ // multiple lines including leading whitespace and the '//' markers.
+ //
+ // FIXME: If we want to handle them correctly, we'll need to adjust
+ // leading whitespace in consecutive lines when changing indentation of
+ // the first line similar to what we do with block comments.
+ StringRef::size_type EscapedNewlinePos = Current.TokenText.find("\\\n");
+ if (EscapedNewlinePos != StringRef::npos) {
+ State.Column =
+ StartColumn +
+ encoding::getCodePointCount(
+ Current.TokenText.substr(0, EscapedNewlinePos), Encoding) +
+ 1;
+ return 0;
+ }
+
Token.reset(new BreakableLineComment(Current, StartColumn,
Line.InPPDirective, Encoding));
} else {
Index: lib/Format/TokenAnnotator.cpp
===================================================================
--- lib/Format/TokenAnnotator.cpp
+++ lib/Format/TokenAnnotator.cpp
@@ -964,8 +964,9 @@
Current->is(tok::string_literal) &&
Current->Previous->isNot(tok::lessless) &&
Current->Previous->Type != TT_InlineASMColon &&
- Current->getNextNonComment() &&
- Current->getNextNonComment()->is(tok::string_literal)) {
+ ((Current->getNextNonComment() &&
+ Current->getNextNonComment()->is(tok::string_literal)) ||
+ (Current->TokenText.find("\\\n") != StringRef::npos))) {
Current->MustBreakBefore = true;
} else {
Current->MustBreakBefore = false;
Index: unittests/Format/FormatTest.cpp
===================================================================
--- unittests/Format/FormatTest.cpp
+++ unittests/Format/FormatTest.cpp
@@ -921,6 +921,32 @@
getLLVMStyleWithColumns(20)));
}
+TEST_F(FormatTest, DontSplitLineCommentsWithEscapedNewlines) {
+ EXPECT_EQ("// aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\\\n"
+ "// aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\\\n"
+ "// aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa",
+ format("// aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\\\n"
+ "// aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\\\n"
+ "// aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa"));
+ EXPECT_EQ("int a; // AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA\\\n"
+ " // AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA\\\n"
+ " // AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA",
+ format("int a; // AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA\\\n"
+ " // AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA\\\n"
+ " // AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA",
+ getLLVMStyleWithColumns(50)));
+ // FIXME: One day we might want to implement adjustment of leading whitespace
+ // of the consecutive lines in this kind of comment:
+ EXPECT_EQ("int\n"
+ "a; // AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA\\\n"
+ " // AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA\\\n"
+ " // AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA",
+ format("int a; // AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA\\\n"
+ " // AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA\\\n"
+ " // AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA",
+ getLLVMStyleWithColumns(49)));
+}
+
TEST_F(FormatTest, PriorityOfCommentBreaking) {
EXPECT_EQ("if (xxx == yyy && // aaaaaaaaaaaa\n"
" // bbbbbbbbb\n"
@@ -2970,6 +2996,21 @@
" \"bbbb\"\n"
" \"cccc\");",
format("aaaa(qqq, \"bbbb\" \"cccc\");", Break));
+ EXPECT_EQ("x = \"a\\\n"
+ "b\\\n"
+ "c\";",
+ format("x = \"a\\\n"
+ "b\\\n"
+ "c\";",
+ NoBreak));
+ EXPECT_EQ("x =\n"
+ " \"a\\\n"
+ "b\\\n"
+ "c\";",
+ format("x = \"a\\\n"
+ "b\\\n"
+ "c\";",
+ Break));
}
TEST_F(FormatTest, AlignsPipes) {
@@ -4991,6 +5032,16 @@
format("#define A \"some text other\";", AlignLeft));
}
+TEST_F(FormatTest, DontSplitStringLiteralsWithEscapedNewlines) {
+ EXPECT_EQ("aaaaaaaaaaa =\n"
+ " \"aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\\\n"
+ " aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\\\n"
+ " aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\";",
+ format("aaaaaaaaaaa = \"aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\\\n"
+ " aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\\\n"
+ " aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\";"));
+}
+
TEST_F(FormatTest, SkipsUnknownStringLiterals) {
EXPECT_EQ(
"u8\"unsupported literal\";",
-------------- next part --------------
A non-text attachment was scrubbed...
Name: D1146.6.patch
Type: text/x-patch
Size: 5905 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20130716/618f1480/attachment.bin>
More information about the cfe-commits
mailing list