From 5b08a09446e2766d61f1a11999011a37e71df225 Mon Sep 17 00:00:00 2001
From: MediaPipe Team <mediapipe-team@google.com>
Date: Tue, 12 Sep 2023 14:57:24 -0700
Subject: [PATCH] Fixes two issues with file handling on windows:

. If UNICODE is set, then win32 functions taking file paths use wide
  (utf-16) strings. For example, FindFirstFile really calls to
  FindFirstFileW, which takes a wchar_t*. This adds support for
  the unicode path.
. SetContents() changes from "w" to "wb". This is necessary as
  windows will do some amount of encoding without "b", which results
  in much different values being written.

PiperOrigin-RevId: 564842317
---
 mediapipe/framework/deps/file_helpers.cc | 45 ++++++++++++++++++------
 1 file changed, 35 insertions(+), 10 deletions(-)

diff --git a/mediapipe/framework/deps/file_helpers.cc b/mediapipe/framework/deps/file_helpers.cc
index 23b50310fa..b5c2bae81f 100644
--- a/mediapipe/framework/deps/file_helpers.cc
+++ b/mediapipe/framework/deps/file_helpers.cc
@@ -17,6 +17,9 @@
 #ifdef _WIN32
 #include <Windows.h>
 #include <direct.h>
+
+#include <codecvt>
+#include <locale>
 #else
 #include <dirent.h>
 #endif  // _WIN32
@@ -86,11 +89,32 @@ class DirectoryListing {
   struct dirent* next_entry_ = nullptr;
 };
 #else
+#if defined(UNICODE)
+using PathString = std::wstring;
+
+PathString Utf8ToNative(const std::string& string) {
+  std::wstring_convert<std::codecvt_utf8_utf16<wchar_t>, wchar_t> converter;
+  return converter.from_bytes(string.data(), string.data() + string.size());
+}
+std::string NativeToUtf8(const PathString& string) {
+  std::wstring_convert<std::codecvt_utf8_utf16<wchar_t>, wchar_t> converter;
+  return converter.to_bytes(string.data(), string.data() + string.size());
+}
+#define FILE_PATH_LITERAL_INTERNAL(x) L##x
+#define FILE_PATH_LITERAL(x) FILE_PATH_LITERAL_INTERNAL(x)
+#else
+using PathString = std::string;
+PathString Utf8ToNative(const std::string& string) { return string; }
+std::string NativeToUtf8(const PathString& string) { return string; }
+#define FILE_PATH_LITERAL_INTERNAL(x) ##x
+#define FILE_PATH_LITERAL(x) FILE_PATH_LITERAL_INTERNAL(x)
+#endif
+
 class DirectoryListing {
  public:
-  explicit DirectoryListing(const std::string& directory) {
-    directory_ = directory;
-    std::string search_string = directory + "\\*.*";
+  explicit DirectoryListing(const std::string& directory)
+      : directory_(Utf8ToNative(directory)) {
+    PathString search_string = directory_ + Utf8ToNative("\\*.*");
     find_handle_ = FindFirstFile(search_string.c_str(), &find_data_);
   }
 
@@ -107,10 +131,10 @@ class DirectoryListing {
   // after the one that is returned, if it exists.
   std::string NextEntry() {
     if (HasNextEntry()) {
-      std::string result =
-          std::string(directory_ + "\\" + find_data_.cFileName);
+      PathString result =
+          directory_ + Utf8ToNative("\\") + PathString(find_data_.cFileName);
       ReadNextEntry();
-      return result;
+      return NativeToUtf8(result);
     } else {
       return std::string();
     }
@@ -119,8 +143,9 @@ class DirectoryListing {
  private:
   void ReadNextEntry() {
     int find_result = FindNextFile(find_handle_, &find_data_);
-    while (find_result != 0 && (std::string(find_data_.cFileName) == "." ||
-                                std::string(find_data_.cFileName) == "..")) {
+    while (find_result != 0 &&
+           (PathString(find_data_.cFileName) == FILE_PATH_LITERAL(".") ||
+            PathString(find_data_.cFileName) == FILE_PATH_LITERAL(".."))) {
       find_result = FindNextFile(find_handle_, &find_data_);
     }
 
@@ -130,7 +155,7 @@ class DirectoryListing {
     }
   }
 
-  std::string directory_;
+  const PathString directory_;
   HANDLE find_handle_ = INVALID_HANDLE_VALUE;
   WIN32_FIND_DATA find_data_;
 };
@@ -162,7 +187,7 @@ absl::Status GetContents(absl::string_view file_name, std::string* output,
 
 absl::Status SetContents(absl::string_view file_name,
                          absl::string_view content) {
-  FILE* fp = fopen(file_name.data(), "w");
+  FILE* fp = fopen(file_name.data(), "wb");
   if (fp == NULL) {
     return mediapipe::InvalidArgumentErrorBuilder(MEDIAPIPE_LOC)
            << "Can't open file: " << file_name;