Add open_file_skipping_BOM function to RC::File

UE4SS-RE · Jun 17, 2024 · a0359bf · a0359bf
1 parent d5eb2f5
commit a0359bf
Show file tree

Hide file tree

Showing 2 changed files with 55 additions and 1 deletion.
diff --git a/deps/first/File/include/File/File.hpp b/deps/first/File/include/File/File.hpp
@@ -12,5 +12,7 @@ namespace RC::File
                           OverwriteExistingFile = OverwriteExistingFile::No,
                           CreateIfNonExistent = CreateIfNonExistent::No) -> Handle;
 
+    RC_FILE_API auto open_file_skip_BOM(const std::wstring& filename) -> StreamType;
+
     RC_FILE_API auto delete_file(const std::filesystem::path& file_path_and_name) -> void;
 } // namespace RC::File
diff --git a/deps/first/File/src/File.cpp b/deps/first/File/src/File.cpp
@@ -1,3 +1,5 @@
+#include <codecvt>
+#include <fstream>
 #include <File/File.hpp>
 
 namespace RC::File
@@ -22,8 +24,58 @@ namespace RC::File
         return construct_handle(file_path_and_name, open_properties);
     }
 
+    // BOM sequences
+    const std::string BOM_UTF8 = "\xEF\xBB\xBF";
+    const std::wstring BOM_UTF16_LE = L"\xFEFF";
+    const std::wstring BOM_UTF16_BE = L"\xFFFE";
+
+    // Function to check for BOM and reopen the file as a wide stream
+    auto open_file_skip_BOM(const std::wstring& filename) -> StreamType
+    {
+        std::ifstream file(filename, std::ios::binary);
+        if (!file.is_open())
+        {
+            throw std::runtime_error("Failed to open the file.");
+        }
+
+        char bom[3] = {0};
+        file.read(bom, 3);
+        if (std::string(bom, 3) == BOM_UTF8)
+        {
+            // UTF-8 BOM detected
+            file.close();
+            StreamType wfile(filename);
+            wfile.seekg(3, std::ios::beg); // Skip BOM
+            return wfile;
+        }
+        else
+        {
+            // Check for wide BOM
+            wchar_t wbom;
+            file.seekg(0, std::ios::beg); // Rewind to start
+            file.read(reinterpret_cast<char*>(&wbom), sizeof(wchar_t));
+            file.close();
+            if (wbom == BOM_UTF16_LE[0] || wbom == BOM_UTF16_BE[0])
+            {
+                StreamType wfile(filename);
+#pragma warning(disable : 4996)
+                std::locale loc(wfile.getloc(), new std::codecvt_utf8_utf16<wchar_t>);
+#pragma warning(default : 4996)
+                wfile.imbue(loc);
+                wfile.seekg(sizeof(wchar_t), std::ios::beg); // Skip BOM
+                return wfile;
+            }
+            else
+            {
+                // No BOM detected
+                file.close();
+                return StreamType(filename);
+            }
+        }
+    }
+
     auto delete_file(const std::filesystem::path& file_path_and_name) -> void
     {
         Handle::FileType::delete_file(file_path_and_name);
     }
-} // namespace RC::File
+} // namespace RC::File