From 9ceafe770d55499674fccbcdf111b194968a72e8 Mon Sep 17 00:00:00 2001
From: Project579 <star579avatar@gmail.com>
Date: Sat, 10 Sep 2022 19:38:29 +0200
Subject: [PATCH] Hard fail on loading BSA with records using unicode paths.

---
 apps/openmw/mwstate/character.cpp    |  2 +-
 components/bsa/bsa_file.cpp          |  2 +-
 components/bsa/bsa_file.hpp          |  2 +-
 components/bsa/compressedbsafile.cpp | 23 +++++++++++++++--------
 components/bsa/compressedbsafile.hpp |  2 +-
 5 files changed, 19 insertions(+), 12 deletions(-)

diff --git a/apps/openmw/mwstate/character.cpp b/apps/openmw/mwstate/character.cpp
index 136adfec89..8e21325c47 100644
--- a/apps/openmw/mwstate/character.cpp
+++ b/apps/openmw/mwstate/character.cpp
@@ -83,7 +83,7 @@ void MWState::Character::addSlot (const ESM::SavedGame& profile)
 }
 
 MWState::Character::Character (std::filesystem::path saves, const std::string& game)
-    : mPath (std::move(saves))
+: mPath (std::move(saves))
 {
     if (!std::filesystem::is_directory (mPath))
     {
diff --git a/components/bsa/bsa_file.cpp b/components/bsa/bsa_file.cpp
index 4bd47a2c92..6a621d91a8 100644
--- a/components/bsa/bsa_file.cpp
+++ b/components/bsa/bsa_file.cpp
@@ -35,7 +35,7 @@ using namespace Bsa;
 
 
 /// Error handling
-[[noreturn]] void BSAFile::fail(const std::string &msg)
+[[noreturn]] void BSAFile::fail(const std::string &msg) const
 {
     throw std::runtime_error("BSA Error: " + msg + "\nArchive: " + Files::pathToUnicodeString(mFilepath));
 }
diff --git a/components/bsa/bsa_file.hpp b/components/bsa/bsa_file.hpp
index 0d1bda6134..b7f3b107fb 100644
--- a/components/bsa/bsa_file.hpp
+++ b/components/bsa/bsa_file.hpp
@@ -90,7 +90,7 @@ protected:
     std::filesystem::path mFilepath;
 
     /// Error handling
-    [[noreturn]] void fail(const std::string &msg);
+    [[noreturn]] void fail(const std::string &msg) const;
 
     /// Read header information from the input source
     virtual void readHeader();
diff --git a/components/bsa/compressedbsafile.cpp b/components/bsa/compressedbsafile.cpp
index ee7204bd9e..ac50ef0c22 100644
--- a/components/bsa/compressedbsafile.cpp
+++ b/components/bsa/compressedbsafile.cpp
@@ -300,6 +300,13 @@ void CompressedBSAFile::readHeader()
 
 CompressedBSAFile::FileRecord CompressedBSAFile::getFileRecord(const std::string& str) const
 {
+    for (const auto c : str)
+    {
+        if (((static_cast<unsigned>(c) >> 7U) & 1U) != 0U) {
+            fail("File record " + str + " contains unicode characters, refusing to load.");
+        }
+    }
+
 #ifdef _WIN32
     const auto& path = str;
 #else
@@ -310,9 +317,9 @@ CompressedBSAFile::FileRecord CompressedBSAFile::getFileRecord(const std::string
     std::replace(path.begin(), path.end(), '\\', '/');
 #endif
 
-    auto p = Files::pathFromUnicodeString(path);
+    const auto p = std::filesystem::path{ path }; // Purposefully damage Unicode strings.
     const auto stem = p.stem();
-    const auto ext = p.extension().u8string();
+    const auto ext = p.extension().string();  // Purposefully damage Unicode strings.
 
     std::uint64_t folderHash = generateHash(p.parent_path(), {});
 
@@ -471,13 +478,13 @@ void CompressedBSAFile::convertCompressedSizesToUncompressed()
     }
 }
 
-std::uint64_t CompressedBSAFile::generateHash(const std::filesystem::path& stem, std::u8string extension)
+std::uint64_t CompressedBSAFile::generateHash(const std::filesystem::path& stem, std::string extension)
 {
     auto str = stem.u8string();
     size_t len = str.length();
     if (len == 0)
         return 0;
-    std::replace(str.begin(), str.end(), u8'/', u8'\\');
+    std::replace(str.begin(), str.end(), '/', '\\');
     Misc::StringUtils::lowerCaseInPlace(str);
     uint64_t result = str[len-1] | (len >= 3 ? (str[len-2] << 8) : 0) | (len << 16) | (str[0] << 24);
     if (len >= 4)
@@ -490,10 +497,10 @@ std::uint64_t CompressedBSAFile::generateHash(const std::filesystem::path& stem,
     if (extension.empty())
         return result;
     Misc::StringUtils::lowerCaseInPlace(extension);
-    if (extension == u8".kf")       result |= 0x80;
-    else if (extension == u8".nif") result |= 0x8000;
-    else if (extension == u8".dds") result |= 0x8080;
-    else if (extension == u8".wav") result |= 0x80000000;
+    if (extension == ".kf")       result |= 0x80;
+    else if (extension == ".nif") result |= 0x8000;
+    else if (extension == ".dds") result |= 0x8080;
+    else if (extension == ".wav") result |= 0x80000000;
     uint32_t hash = 0;
     for (const auto &c : extension)
         hash = hash * 0x1003f + c;
diff --git a/components/bsa/compressedbsafile.hpp b/components/bsa/compressedbsafile.hpp
index 71727bfade..bad8551f09 100644
--- a/components/bsa/compressedbsafile.hpp
+++ b/components/bsa/compressedbsafile.hpp
@@ -83,7 +83,7 @@ namespace Bsa
         //mFiles used by OpenMW will contain uncompressed file sizes
         void convertCompressedSizesToUncompressed();
         /// \brief Normalizes given filename or folder and generates format-compatible hash. See https://en.uesp.net/wiki/Tes4Mod:Hash_Calculation.
-        static std::uint64_t generateHash(const std::filesystem::path& stem, std::u8string extension) ;
+        static std::uint64_t generateHash(const std::filesystem::path& stem, std::string extension) ;
         Files::IStreamPtr getFile(const FileRecord& fileRecord);
     public:
         using BSAFile::open;