[llvm] [llvm][support] Refactor symlink handling and add readlink (PR #184256)

[Michael Spencer via llvm-commits]

================
@@ -1649,6 +1683,96 @@ std::error_code real_path(const Twine &path, SmallVectorImpl<char> &dest,
   return std::error_code();
 }
 
+// This struct is normally only available in the Windows Driver Kit (WDK)
+// headers, not in the standard Windows SDK.
+struct LLVM_REPARSE_DATA_BUFFER {
+  unsigned long ReparseTag;
+  unsigned short ReparseDataLength;
+  unsigned short Reserved;
+  union {
+    struct {
+      unsigned short SubstituteNameOffset;
+      unsigned short SubstituteNameLength;
+      unsigned short PrintNameOffset;
+      unsigned short PrintNameLength;
+      unsigned long Flags;
+      wchar_t PathBuffer[1];
+    } SymbolicLinkReparseBuffer;
+    struct {
+      unsigned short SubstituteNameOffset;
+      unsigned short SubstituteNameLength;
+      unsigned short PrintNameOffset;
+      unsigned short PrintNameLength;
+      wchar_t PathBuffer[1];
+    } MountPointReparseBuffer;
+    struct {
+      unsigned char DataBuffer[1];
+    } GenericReparseBuffer;
+  };
+};
+
+std::error_code readlink(const Twine &path, SmallVectorImpl<char> &dest) {
+  dest.clear();
+
+  SmallVector<wchar_t, 128> PathUTF16;
+  if (std::error_code EC = widenPath(path, PathUTF16))
+    return EC;
+
+  // Open the symlink without following it.
+  ScopedFileHandle H(::CreateFileW(
+      c_str(PathUTF16), FILE_READ_ATTRIBUTES,
+      FILE_SHARE_READ | FILE_SHARE_WRITE | FILE_SHARE_DELETE, NULL,
+      OPEN_EXISTING, FILE_FLAG_BACKUP_SEMANTICS | FILE_FLAG_OPEN_REPARSE_POINT,
+      NULL));
+  if (!H)
+    return mapWindowsError(::GetLastError());
+
+  // Read the reparse point data.
+  union {
+    LLVM_REPARSE_DATA_BUFFER RDB;
+    char Buffer[MAXIMUM_REPARSE_DATA_BUFFER_SIZE];
+  };
+  DWORD BytesReturned;
+  if (!::DeviceIoControl(H, FSCTL_GET_REPARSE_POINT, NULL, 0, &RDB,
+                         sizeof(Buffer), &BytesReturned, NULL)) {
+    DWORD Err = ::GetLastError();
+    if (Err == ERROR_NOT_A_REPARSE_POINT)
+      return make_error_code(errc::invalid_argument);
+    return mapWindowsError(Err);
+  }
+
+  if (RDB.ReparseTag != IO_REPARSE_TAG_SYMLINK)
+    return make_error_code(errc::invalid_argument);
+
+  const auto &SLB = RDB.SymbolicLinkReparseBuffer;
+  size_t PathBufOffset =
+      offsetof(LLVM_REPARSE_DATA_BUFFER, SymbolicLinkReparseBuffer.PathBuffer);
+
+  // Prefer PrintName (user-friendly, e.g. "C:\foo") over SubstituteName
+  // (NT-internal, e.g. "\??\C:\foo").
+  USHORT NameOffset, NameLength;
+  if (SLB.PrintNameLength != 0) {
+    NameOffset = SLB.PrintNameOffset;
+    NameLength = SLB.PrintNameLength;
+  } else {
+    NameOffset = SLB.SubstituteNameOffset;
+    NameLength = SLB.SubstituteNameLength;
+  }
+
+  // Validate that the returned data is large enough to contain the name.
+  if (PathBufOffset + NameOffset + NameLength > BytesReturned)
+    return make_error_code(errc::invalid_argument);
+
+  const wchar_t *Target = SLB.PathBuffer + NameOffset / sizeof(wchar_t);
+  USHORT TargetLen = NameLength / sizeof(wchar_t);
+
+  if (std::error_code EC = UTF16ToUTF8(Target, TargetLen, dest))
+    return EC;
+
+  llvm::sys::path::make_preferred(dest);
----------------
Bigcheese wrote:

I originally was doing this because we do it elsewhere, but we're actually kinda inconsistent about it. I don't feel too strongly either way; path separators in LLVM/Clang on Windows end up as a mix a lot of the time anyway.

I'll remove it to be consistent with Unix and not remove information.

https://github.com/llvm/llvm-project/pull/184256