[clang] [Clang] Warn on backslash-newline-EOF (PR #97585)
Mital Ashok via cfe-commits
cfe-commits at lists.llvm.org
Thu Jul 4 03:30:35 PDT 2024
https://github.com/MitalAshok updated https://github.com/llvm/llvm-project/pull/97585
>From 8af656659b79d76c971b01f1f4c14dc7315565b8 Mon Sep 17 00:00:00 2001
From: Mital Ashok <mital at mitalashok.co.uk>
Date: Fri, 21 Jun 2024 18:55:38 +0100
Subject: [PATCH 1/4] [Clang] Warn on backslash-newline-EOF
---
clang/docs/ReleaseNotes.rst | 2 +
.../include/clang/Basic/DiagnosticLexKinds.td | 1 +
clang/lib/Lex/Lexer.cpp | 39 +++++++++++++++++--
clang/test/CXX/drs/cwg16xx.cpp | 9 +++++
clang/test/CXX/drs/cwg2747.cpp | 11 ++++++
clang/test/CXX/drs/cwg27xx.cpp | 2 +
.../test/Preprocessor/backslash_newline_eof.c | 12 ++++++
.../Preprocessor/backslash_without_newline.c | 8 ++++
.../Preprocessor/backslash_without_newline.h | 4 ++
clang/www/cxx_dr_status.html | 4 +-
10 files changed, 87 insertions(+), 5 deletions(-)
create mode 100644 clang/test/CXX/drs/cwg2747.cpp
create mode 100644 clang/test/Preprocessor/backslash_newline_eof.c
create mode 100644 clang/test/Preprocessor/backslash_without_newline.c
create mode 100644 clang/test/Preprocessor/backslash_without_newline.h
diff --git a/clang/docs/ReleaseNotes.rst b/clang/docs/ReleaseNotes.rst
index f40fd1cd145bb0..7c0ac3a504f982 100644
--- a/clang/docs/ReleaseNotes.rst
+++ b/clang/docs/ReleaseNotes.rst
@@ -647,6 +647,8 @@ Improvements to Clang's diagnostics
- Clang now shows implicit deduction guides when diagnosing overload resolution failure. #GH92393.
+- Clang now emits ``-Wnewline-eof`` when the last newline is deleted by a preceding backslash.
+
Improvements to Clang's time-trace
----------------------------------
diff --git a/clang/include/clang/Basic/DiagnosticLexKinds.td b/clang/include/clang/Basic/DiagnosticLexKinds.td
index 12d7b8c0205ee9..e6b2c1385944c7 100644
--- a/clang/include/clang/Basic/DiagnosticLexKinds.td
+++ b/clang/include/clang/Basic/DiagnosticLexKinds.td
@@ -56,6 +56,7 @@ def ext_no_newline_eof : Extension<"no newline at end of file">,
InGroup<NewlineEOF>;
def warn_no_newline_eof : Warning<"no newline at end of file">,
InGroup<NewlineEOF>, DefaultIgnore;
+def note_backslash_newline_eof : Note<"last newline deleted by splice here">;
def warn_cxx98_compat_no_newline_eof : Warning<
"C++98 requires newline at end of file">,
diff --git a/clang/lib/Lex/Lexer.cpp b/clang/lib/Lex/Lexer.cpp
index e59c7805b38623..0e540834b473ba 100644
--- a/clang/lib/Lex/Lexer.cpp
+++ b/clang/lib/Lex/Lexer.cpp
@@ -3165,7 +3165,17 @@ bool Lexer::LexEndOfFile(Token &Result, const char *CurPtr) {
// C99 5.1.1.2p2: If the file is non-empty and didn't end in a newline, issue
// a pedwarn.
- if (CurPtr != BufferStart && (CurPtr[-1] != '\n' && CurPtr[-1] != '\r')) {
+ if (CurPtr != BufferStart) {
+ StringRef LastNewline;
+ if (CurPtr[-1] == '\r' || CurPtr[-1] == '\n') {
+ LastNewline = StringRef(CurPtr - 1, 1);
+ if (CurPtr - 1 != BufferStart && CurPtr[-2] != CurPtr[-1] &&
+ (CurPtr[-2] == '\r' || CurPtr[-2] == '\n')) {
+ // \r\n or \n\r is one newline
+ LastNewline = StringRef(CurPtr - 2, 2);
+ }
+ }
+
DiagnosticsEngine &Diags = PP->getDiagnostics();
SourceLocation EndLoc = getSourceLocation(BufferEnd);
unsigned DiagID;
@@ -3183,8 +3193,31 @@ bool Lexer::LexEndOfFile(Token &Result, const char *CurPtr) {
DiagID = diag::ext_no_newline_eof;
}
- Diag(BufferEnd, DiagID)
- << FixItHint::CreateInsertion(EndLoc, "\n");
+ if (LastNewline.empty()) {
+ Diag(BufferEnd, DiagID) << FixItHint::CreateInsertion(EndLoc, "\n");
+ } else {
+ // While the file physically ends in a newline, the previous
+ // line might have ended in a splice, so it would be deleted
+ const char *LastSpliceLocation = LastNewline.data();
+ while (LastSpliceLocation != BufferStart &&
+ isHorizontalWhitespace(*--LastSpliceLocation))
+ ;
+
+ bool LastIsSplice = *LastSpliceLocation == '\\';
+ if (*LastSpliceLocation == '/' && LangOpts.Trigraphs)
+ // Check for "??/" trigraph for "\"
+ LastIsSplice =
+ LastSpliceLocation != BufferStart && *--LastSpliceLocation == '?' &&
+ LastSpliceLocation != BufferStart && *--LastSpliceLocation == '?';
+
+ if (LastIsSplice) {
+ PP->Diag(getSourceLocation(LastNewline.data(), LastNewline.size()),
+ DiagID);
+ Diag(LastSpliceLocation, diag::note_backslash_newline_eof)
+ << FixItHint::CreateRemoval(getSourceLocation(
+ LastSpliceLocation, *LastSpliceLocation == '\\' ? 1 : 3));
+ }
+ }
}
BufferPtr = CurPtr;
diff --git a/clang/test/CXX/drs/cwg16xx.cpp b/clang/test/CXX/drs/cwg16xx.cpp
index cf6b45ceabf2cc..dca941fa30624f 100644
--- a/clang/test/CXX/drs/cwg16xx.cpp
+++ b/clang/test/CXX/drs/cwg16xx.cpp
@@ -536,3 +536,12 @@ namespace cwg1696 { // cwg1696: 7
};
#endif
}
+
+// cwg1698: yes
+// This file intentionally does not end in a newline
+// to facilitate the CWG1698 test
+
+// cxx98-error at +3 {{no newline at end of file}}
+// expected-error at +2 {{expected unqualified-id}}
+
+\
\ No newline at end of file
diff --git a/clang/test/CXX/drs/cwg2747.cpp b/clang/test/CXX/drs/cwg2747.cpp
new file mode 100644
index 00000000000000..3f971452c5193f
--- /dev/null
+++ b/clang/test/CXX/drs/cwg2747.cpp
@@ -0,0 +1,11 @@
+// RUN: %clang_cc1 -std=c++11 -pedantic-errors -verify=expected %s -E | FileCheck %s --strict-whitespace --allow-empty
+
+// expected-no-diagnostics
+// cwg2747: yes
+
+// Check that a newline is still added even though there is a
+// physical newline at the end of the file (which is spliced)
+// CHECK: int x;{{$[[:space:]]^}}int y;int z;{{$[[:space:]]^$}}
+int x;
+int y;\
+int z;\
diff --git a/clang/test/CXX/drs/cwg27xx.cpp b/clang/test/CXX/drs/cwg27xx.cpp
index 406c8ea41f3b2f..6e47af6333569d 100644
--- a/clang/test/CXX/drs/cwg27xx.cpp
+++ b/clang/test/CXX/drs/cwg27xx.cpp
@@ -18,6 +18,8 @@ void f(B b) {
struct D : B {};
} // namespace cwg2718
+// cwg2747 is in cwg2747.cpp
+
namespace cwg2759 { // cwg2759: 19
#if __cplusplus >= 201103L
diff --git a/clang/test/Preprocessor/backslash_newline_eof.c b/clang/test/Preprocessor/backslash_newline_eof.c
new file mode 100644
index 00000000000000..b41bc92a425144
--- /dev/null
+++ b/clang/test/Preprocessor/backslash_newline_eof.c
@@ -0,0 +1,12 @@
+// RUN: %clang_cc1 -x c -fsyntax-only -pedantic -verify %s
+// RUN: %clang_cc1 -x c++ -fsyntax-only -std=c++98 -pedantic -verify %s
+// RUN: %clang_cc1 -x c++ -fsyntax-only -std=c++11 -Wc++98-compat-pedantic -verify=cxx11-compat %s
+// RUN: %clang_cc1 -x c++ -fsyntax-only -std=c++11 -verify=cxx11 %s
+
+// cxx11-no-diagnostics
+
+// expected-warning at +4 {{no newline at end of file}}
+// expected-note at +3 {{last newline deleted by splice here}}
+// cxx11-compat-warning at +2 {{C++98 requires newline at end of file}}
+// cxx11-compat-note at +1 {{last newline deleted by splice here}}
+int x; \
diff --git a/clang/test/Preprocessor/backslash_without_newline.c b/clang/test/Preprocessor/backslash_without_newline.c
new file mode 100644
index 00000000000000..84be4e04b54adc
--- /dev/null
+++ b/clang/test/Preprocessor/backslash_without_newline.c
@@ -0,0 +1,8 @@
+// RUN: %clang_cc1 -x c -E -o - %s | FileCheck %s
+// RUN: %clang_cc1 -x c++ -E -o - %s | FileCheck %s
+// RUN: %clang_cc1 -x c++ -E -o - %s | FileCheck %s
+
+#include "./backslash_without_newline.h"
+
+// CHECK: A B \ C
+A BACKSLASH_WITH_NEWLINE B BACKSLASH_WITHOUT_NEWLINE C
diff --git a/clang/test/Preprocessor/backslash_without_newline.h b/clang/test/Preprocessor/backslash_without_newline.h
new file mode 100644
index 00000000000000..db21184332c3b2
--- /dev/null
+++ b/clang/test/Preprocessor/backslash_without_newline.h
@@ -0,0 +1,4 @@
+#define BACKSLASH_WITH_NEWLINE \
+
+// This file intentionally does not end with a newline.
+#define BACKSLASH_WITHOUT_NEWLINE \
\ No newline at end of file
diff --git a/clang/www/cxx_dr_status.html b/clang/www/cxx_dr_status.html
index 937f67981e2963..2da096b746611e 100755
--- a/clang/www/cxx_dr_status.html
+++ b/clang/www/cxx_dr_status.html
@@ -10003,7 +10003,7 @@ <h2 id="cxxdr">C++ defect report implementation status</h2>
<td><a href="https://cplusplus.github.io/CWG/issues/1698.html">1698</a></td>
<td>DRWP</td>
<td>Files ending in <TT>\</TT></td>
- <td class="unknown" align="center">Unknown</td>
+ <td class="full" align="center">Yes</td>
</tr>
<tr class="open" id="1699">
<td><a href="https://cplusplus.github.io/CWG/issues/1699.html">1699</a></td>
@@ -16297,7 +16297,7 @@ <h2 id="cxxdr">C++ defect report implementation status</h2>
<td><a href="https://cplusplus.github.io/CWG/issues/2747.html">2747</a></td>
<td>DRWP</td>
<td>Cannot depend on an already-deleted splice</td>
- <td class="unknown" align="center">Unknown</td>
+ <td class="full" align="center">Yes</td>
</tr>
<tr id="2748">
<td><a href="https://cplusplus.github.io/CWG/issues/2748.html">2748</a></td>
>From 8b78310bf05e3d7459e19eb9e44d4cc805a71987 Mon Sep 17 00:00:00 2001
From: Mital Ashok <mital at mitalashok.co.uk>
Date: Thu, 4 Jul 2024 10:37:31 +0100
Subject: [PATCH 2/4] Add tests about trigraphs and trailing spaces
---
.../Lexer/backslash-trigraph_newline_eof.c | 18 ++++++++++++++++++
.../backslash-trigraph_space-ext-newline_eof.c | 18 ++++++++++++++++++
clang/test/Lexer/backslash_newline_eof.c | 14 ++++++++++++++
.../Lexer/backslash_space-ext-newline_eof.c | 12 ++++++++++++
.../test/Preprocessor/backslash_newline_eof.c | 12 ------------
5 files changed, 62 insertions(+), 12 deletions(-)
create mode 100644 clang/test/Lexer/backslash-trigraph_newline_eof.c
create mode 100644 clang/test/Lexer/backslash-trigraph_space-ext-newline_eof.c
create mode 100644 clang/test/Lexer/backslash_newline_eof.c
create mode 100644 clang/test/Lexer/backslash_space-ext-newline_eof.c
delete mode 100644 clang/test/Preprocessor/backslash_newline_eof.c
diff --git a/clang/test/Lexer/backslash-trigraph_newline_eof.c b/clang/test/Lexer/backslash-trigraph_newline_eof.c
new file mode 100644
index 00000000000000..f066168f3a3f72
--- /dev/null
+++ b/clang/test/Lexer/backslash-trigraph_newline_eof.c
@@ -0,0 +1,18 @@
+// RUN: %clang_cc1 -x c -std=c18 -fsyntax-only -pedantic -verify=expected,trigraphs,no-newline %s
+// RUN: %clang_cc1 -x c -std=c23 -fsyntax-only -pedantic -verify=expected,no-trigraphs,c23 %s
+// RUN: %clang_cc1 -x c++ -fsyntax-only -std=c++98 -pedantic -verify=expected,trigraphs,no-newline %s
+// RUN: %clang_cc1 -x c++ -fsyntax-only -std=c++11 -Wc++98-compat-pedantic -verify=expected,trigraphs,no-newline-compat %s
+// RUN: %clang_cc1 -x c++ -fsyntax-only -std=c++11 -verify=expected,trigraphs %s
+// RUN: %clang_cc1 -x c++ -fsyntax-only -std=c++17 -verify=expected,no-trigraphs,cxx17 %s
+
+// trigraphs-warning at +10 {{trigraph converted to '\' character}}
+// no-newline-warning at +9 {{no newline at end of file}}
+// no-newline-note at +8 {{last newline deleted by splice here}}
+// no-newline-compat-warning at +7 {{C++98 requires newline at end of file}}
+// no-newline-compat-note at +6 {{last newline deleted by splice here}}
+// no-trigraphs-warning at +5 {{trigraph ignored}}
+// c23-error at +4 {{expected identifier or '('}}
+// cxx17-error at +3 {{expected unqualified-id}}
+
+
+int x; ??/
diff --git a/clang/test/Lexer/backslash-trigraph_space-ext-newline_eof.c b/clang/test/Lexer/backslash-trigraph_space-ext-newline_eof.c
new file mode 100644
index 00000000000000..fdf51000bd255e
--- /dev/null
+++ b/clang/test/Lexer/backslash-trigraph_space-ext-newline_eof.c
@@ -0,0 +1,18 @@
+// RUN: %clang_cc1 -x c -std=c18 -fsyntax-only -pedantic -verify=expected,trigraphs,no-newline %s
+// RUN: %clang_cc1 -x c -std=c23 -fsyntax-only -pedantic -verify=expected,no-trigraphs,c23 %s
+// RUN: %clang_cc1 -x c++ -fsyntax-only -std=c++98 -pedantic -verify=expected,trigraphs,no-newline %s
+// RUN: %clang_cc1 -x c++ -fsyntax-only -std=c++11 -Wc++98-compat-pedantic -verify=expected,trigraphs,no-newline-compat %s
+// RUN: %clang_cc1 -x c++ -fsyntax-only -std=c++11 -verify=expected,trigraphs %s
+// RUN: %clang_cc1 -x c++ -fsyntax-only -std=c++17 -verify=expected,no-trigraphs,cxx17 %s
+
+// trigraphs-warning at +10 {{trigraph converted to '\' character}}
+// no-newline-warning at +9 {{no newline at end of file}}
+// no-newline-note at +8 {{last newline deleted by splice here}}
+// no-newline-compat-warning at +7 {{C++98 requires newline at end of file}}
+// no-newline-compat-note at +6 {{last newline deleted by splice here}}
+// no-trigraphs-warning at +5 {{trigraph ignored}}
+// c23-error at +4 {{expected identifier or '('}}
+// cxx17-error at +3 {{expected unqualified-id}}
+// trigraphs-warning at +2 {{backslash and newline separated by space}}
+// The next line intentionally has a trailing tab character.
+int x; ??/
diff --git a/clang/test/Lexer/backslash_newline_eof.c b/clang/test/Lexer/backslash_newline_eof.c
new file mode 100644
index 00000000000000..0288efaf10bd43
--- /dev/null
+++ b/clang/test/Lexer/backslash_newline_eof.c
@@ -0,0 +1,14 @@
+// RUN: %clang_cc1 -x c -fsyntax-only -pedantic -verify=expected,no-newline %s
+// RUN: %clang_cc1 -x c++ -fsyntax-only -std=c++98 -pedantic -verify=expected,no-newline %s
+// RUN: %clang_cc1 -x c++ -fsyntax-only -std=c++11 -Wc++98-compat-pedantic -verify=expected,no-newline-compat %s
+// RUN: %clang_cc1 -x c++ -fsyntax-only -std=c++11 -verify=expected,cxx11 %s
+
+// cxx11-no-diagnostics
+
+// no-newline-warning at +6 {{no newline at end of file}}
+// no-newline-note at +5 {{last newline deleted by splice here}}
+// no-newline-compat-warning at +4 {{C++98 requires newline at end of file}}
+// no-newline-compat-note at +3 {{last newline deleted by splice here}}
+
+
+int x; \
diff --git a/clang/test/Lexer/backslash_space-ext-newline_eof.c b/clang/test/Lexer/backslash_space-ext-newline_eof.c
new file mode 100644
index 00000000000000..d9cb8c38b144f8
--- /dev/null
+++ b/clang/test/Lexer/backslash_space-ext-newline_eof.c
@@ -0,0 +1,12 @@
+// RUN: %clang_cc1 -x c -fsyntax-only -pedantic -verify=expected,no-newline %s
+// RUN: %clang_cc1 -x c++ -fsyntax-only -std=c++98 -pedantic -verify=expected,no-newline %s
+// RUN: %clang_cc1 -x c++ -fsyntax-only -std=c++11 -Wc++98-compat-pedantic -verify=expected,no-newline-compat %s
+// RUN: %clang_cc1 -x c++ -fsyntax-only -std=c++11 -verify=expected,cxx11 %s
+
+// no-newline-warning at +6 {{no newline at end of file}}
+// no-newline-note at +5 {{last newline deleted by splice here}}
+// no-newline-compat-warning at +4 {{C++98 requires newline at end of file}}
+// no-newline-compat-note at +3 {{last newline deleted by splice here}}
+// expected-warning at +2 {{backslash and newline separated by space}}
+// The next line intentionally has a trailing tab character.
+int x; \
diff --git a/clang/test/Preprocessor/backslash_newline_eof.c b/clang/test/Preprocessor/backslash_newline_eof.c
deleted file mode 100644
index b41bc92a425144..00000000000000
--- a/clang/test/Preprocessor/backslash_newline_eof.c
+++ /dev/null
@@ -1,12 +0,0 @@
-// RUN: %clang_cc1 -x c -fsyntax-only -pedantic -verify %s
-// RUN: %clang_cc1 -x c++ -fsyntax-only -std=c++98 -pedantic -verify %s
-// RUN: %clang_cc1 -x c++ -fsyntax-only -std=c++11 -Wc++98-compat-pedantic -verify=cxx11-compat %s
-// RUN: %clang_cc1 -x c++ -fsyntax-only -std=c++11 -verify=cxx11 %s
-
-// cxx11-no-diagnostics
-
-// expected-warning at +4 {{no newline at end of file}}
-// expected-note at +3 {{last newline deleted by splice here}}
-// cxx11-compat-warning at +2 {{C++98 requires newline at end of file}}
-// cxx11-compat-note at +1 {{last newline deleted by splice here}}
-int x; \
>From 1e8ca82a35e42fe8979bf7f86b76f71ea44d12a9 Mon Sep 17 00:00:00 2001
From: Mital Ashok <mital at mitalashok.co.uk>
Date: Thu, 4 Jul 2024 11:17:03 +0100
Subject: [PATCH 3/4] Hopefully this is more readable
---
clang/lib/Lex/Lexer.cpp | 36 ++++++++++++++++++++----------------
1 file changed, 20 insertions(+), 16 deletions(-)
diff --git a/clang/lib/Lex/Lexer.cpp b/clang/lib/Lex/Lexer.cpp
index 0e540834b473ba..a5b3eaabb3a4dc 100644
--- a/clang/lib/Lex/Lexer.cpp
+++ b/clang/lib/Lex/Lexer.cpp
@@ -3198,24 +3198,28 @@ bool Lexer::LexEndOfFile(Token &Result, const char *CurPtr) {
} else {
// While the file physically ends in a newline, the previous
// line might have ended in a splice, so it would be deleted
- const char *LastSpliceLocation = LastNewline.data();
- while (LastSpliceLocation != BufferStart &&
- isHorizontalWhitespace(*--LastSpliceLocation))
- ;
-
- bool LastIsSplice = *LastSpliceLocation == '\\';
- if (*LastSpliceLocation == '/' && LangOpts.Trigraphs)
- // Check for "??/" trigraph for "\"
- LastIsSplice =
- LastSpliceLocation != BufferStart && *--LastSpliceLocation == '?' &&
- LastSpliceLocation != BufferStart && *--LastSpliceLocation == '?';
-
- if (LastIsSplice) {
+ StringRef WithoutLastNewline =
+ StringRef(BufferStart, LastNewline.data() - BufferStart);
+ while (!WithoutLastNewline.empty()) {
+ if (isHorizontalWhitespace(WithoutLastNewline.back())) {
+ WithoutLastNewline = WithoutLastNewline.drop_back();
+ } else {
+ break;
+ }
+ }
+
+ if (WithoutLastNewline.ends_with('\\') ||
+ (LangOpts.Trigraphs && WithoutLastNewline.ends_with("??"
+ "/"))) {
PP->Diag(getSourceLocation(LastNewline.data(), LastNewline.size()),
DiagID);
- Diag(LastSpliceLocation, diag::note_backslash_newline_eof)
- << FixItHint::CreateRemoval(getSourceLocation(
- LastSpliceLocation, *LastSpliceLocation == '\\' ? 1 : 3));
+ std::size_t SpliceSize = WithoutLastNewline.back() == '\\' ? 1 : 3;
+ SourceLocation LastSpliceLocation =
+ getSourceLocation(WithoutLastNewline.data() +
+ (WithoutLastNewline.size() - SpliceSize),
+ SpliceSize);
+ PP->Diag(LastSpliceLocation, diag::note_backslash_newline_eof)
+ << FixItHint::CreateRemoval(LastSpliceLocation);
}
}
}
>From eca9a7a36ecf7dfa487bc74a9dda2137bb35b46a Mon Sep 17 00:00:00 2001
From: Mital Ashok <mital at mitalashok.co.uk>
Date: Thu, 4 Jul 2024 11:30:21 +0100
Subject: [PATCH 4/4] Found related GitHub issue
---
clang/docs/ReleaseNotes.rst | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/clang/docs/ReleaseNotes.rst b/clang/docs/ReleaseNotes.rst
index 7c0ac3a504f982..8493cc838013d5 100644
--- a/clang/docs/ReleaseNotes.rst
+++ b/clang/docs/ReleaseNotes.rst
@@ -647,7 +647,7 @@ Improvements to Clang's diagnostics
- Clang now shows implicit deduction guides when diagnosing overload resolution failure. #GH92393.
-- Clang now emits ``-Wnewline-eof`` when the last newline is deleted by a preceding backslash.
+- Clang now emits ``-Wnewline-eof`` when the last newline is deleted by a preceding backslash. #GH41571.
Improvements to Clang's time-trace
----------------------------------
More information about the cfe-commits
mailing list