[split-file] Respect input file's line endings

[beanz]

Summary

This change adds support for split-file to respect the line ending style
of the input file. This enables split-file to work as expected on
Windows with input files containing CRLF line endings.

The test files added along with this change mirror the existing basic
tests, but are forced to contain CRLF line endings via git attributes.
This will result in the tests always containing CRLF line endings when
checked out regardless of the user's OS.

[MaskRay]

I think llvm-project tests normally should all use '\n', but this change may make the utility more useful outside llvm-project, so looks good.

llvm/test/tools/split-file/basic.crlf.test

1

We do not these many tests. One crlf file is sufficient.

[beanz]

In D117897#3261865, @MaskRay wrote:

I think llvm-project tests normally should all use '\n', but this change may make the utility more useful outside llvm-project, so looks good.

The fact that they are required to be LF ended was intended to be temporary while the issues with CRLF got worked out (see the note: https://llvm.org/docs/GettingStarted.html#checkout-llvm-from-git).

This is me trying to work out the issues :)

[beanz]

Updates based on PR feedback from @MaskRay

[MaskRay]

Looks great!

llvm/.gitattributes

 # them as always having crlf line endings
 test/MC/AsmParser/preserve-comments-crlf.s text eol=crlf
 test/tools/llvm-mca/X86/directives-handle-crlf.s text eol=crlf
 
 # These files must have LF line endings because the test requires exact matching
 test/tools/llvm-strings/radix.test text eol=lf
 test/tools/split-file/basic.test text eol=lf
 test/tools/split-file/Inputs/basic-*.txt eol=lf
+test/tools/split-file/basic.crlf.test text eol=crlf
+test/tools/split-file/Inputs/basic-*.crlf eol=crlf

llvm/test/tools/split-file/Inputs/basic-aa.crlf

+
+aa

llvm/test/tools/split-file/Inputs/basic-bb.crlf

+
+
+
+; Comments are preserved.
+bb
+

llvm/test/tools/split-file/Inputs/basic-cc.crlf

+
+
+
+
+
+
+
+cc

llvm/test/tools/split-file/basic.crlf.test

+#--- aa
+aa
+;--- bb
+; Comments are preserved.
+bb
+
+//--- subdir/cc
+cc
+//--- end
+
+# RUN: rm -rf %t
+# RUN: split-file --leading-lines %s %t
+# RUN: diff %S/Inputs/basic-aa.crlf %t/aa
+# RUN: diff %S/Inputs/basic-bb.crlf %t/bb
+# RUN: diff %S/Inputs/basic-cc.crlf %t/subdir/cc
+# RUN: FileCheck %s --check-prefix=END < %t/end
+
+## Can be called on a non-empty directory.
+# RUN: split-file --leading-lines %s %t
+# RUN: diff %S/Inputs/basic-aa.crlf %t/aa
+
+## Test that we will delete the output if it is a file, so that we can create
+## a directory.
+# RUN: rm -rf %t && touch %t
+# RUN: split-file --leading-lines %s %t
+# RUN: diff %S/Inputs/basic-aa.crlf %t/aa
+
+# END: RUN: split-file %s %t
+
+# RUN: not %split-file 2>&1 | FileCheck %s --check-prefix=NO_INPUT
+
+# NO_INPUT: split-file: error: input filename is not specified
+
+# RUN: not %split-file %s '' 2>&1 | FileCheck %s --check-prefix=NO_OUTPUT
+
+# NO_OUTPUT: split-file: error: output directory is not specified
+
+# RUN: not %split-file %S/Inputs/basic-aa.crlf %t 2>&1 | FileCheck %s --check-prefix=NOT_EXIST
+
+# NOT_EXIST: split-file: error: {{.*}}: no part separator was found

llvm/tools/split-file/split-file.cpp

struct Part {
   const char *end = nullptr;
   int64_t leadingLines = 0;
 };
 } // namespace
 
 static int handle(MemoryBuffer &inputBuf, StringRef input) {
   DenseMap<StringRef, Part> partToBegin;
   StringRef lastPart, separator;
+  StringRef EOL = inputBuf.getBuffer().detectEOL();
   for (line_iterator i(inputBuf, /*SkipBlanks=*/false, '\0'); !i.is_at_eof();) {
     const int64_t lineNo = i.line_number();
     const StringRef line = *i++;
     const size_t markerLen = line.startswith("//") ? 6 : 5;
     if (!(line.size() >= markerLen &&
           line.substr(markerLen - 4).startswith("--- ")))
       continue;
     separator = line.substr(0, markerLen);
if (ec)
       fatal(input, ec.message());
     auto f = std::make_unique<ToolOutputFile>(partPath.str(), ec,
                                               llvm::sys::fs::OF_None);
     if (!f)
       fatal(input, ec.message());
 
     Part &part = keyValue.second;
     for (int64_t i = 0; i != part.leadingLines; ++i)
-      (*f).os().write('\n');
+      (*f).os() << EOL;
     if (part.begin)
       (*f).os().write(part.begin, part.end - part.begin);
     outputFiles.push_back(std::move(f));
   }
 
   for (std::unique_ptr<ToolOutputFile> &outputFile : outputFiles)
     outputFile->keep();
   return 0;