Merge branch 'esm4_zero_terminated_strings' into 'master'

Fix reading array of zero terminated strings (ESM4 NIFZ and KFFZ sub records)

See merge request OpenMW/openmw!2403

apps/esmtool/tes4.cpp

@@ -80,6 +80,45 @@ namespace EsmTool
         template <class T>
         constexpr bool hasModel = HasModel<T>::value;
 
+        template <class T, class = std::void_t<>>
+        struct HasNif : std::false_type {};
+
+        template <class T>
+        struct HasNif<T, std::void_t<decltype(T::mNif)>> : std::true_type {};
+
+        template <class T>
+        constexpr bool hasNif = HasNif<T>::value;
+
+        template <class T, class = std::void_t<>>
+        struct HasKf : std::false_type {};
+
+        template <class T>
+        struct HasKf<T, std::void_t<decltype(T::mKf)>> : std::true_type {};
+
+        template <class T>
+        constexpr bool hasKf = HasKf<T>::value;
+
+        template <class T>
+        struct WriteArray
+        {
+            std::string_view mPrefix;
+            const T& mValue;
+
+            explicit WriteArray(std::string_view prefix, const T& value)
+                : mPrefix(prefix)
+                , mValue(value)
+            {
+            }
+        };
+
+        template <class T>
+        std::ostream& operator<<(std::ostream& stream, const WriteArray<T>& write)
+        {
+            for (const auto& value : write.mValue)
+                stream << write.mPrefix << value;
+            return stream;
+        }
+
         template <class T>
         void readTypedRecord(const Params& params, ESM4::Reader& reader)
         {
@@ -100,6 +139,10 @@ namespace EsmTool
                 std::cout << "\n  EditorId: " << value.mEditorId;
             if constexpr (hasModel<T>)
                 std::cout << "\n  Model: " << value.mModel;
+            if constexpr (hasNif<T>)
+                std::cout << "\n  Nif:" << WriteArray("\n  - ", value.mNif);
+            if constexpr (hasKf<T>)
+                std::cout << "\n  Kf:" << WriteArray("\n  - ", value.mKf);
             std::cout << '\n';
         }
 

components/esm4/loadcrea.cpp

@@ -121,15 +121,8 @@ void ESM4::Creature::load(ESM4::Reader& reader)
             case ESM4::SUB_NAM1: reader.getZString(mBloodDecal); break;
             case ESM4::SUB_NIFZ:
             {
-                std::string str;
+                if (!reader.getZeroTerminatedStringArray(mNif))
-                if (!reader.getZString(str))
                     throw std::runtime_error ("CREA NIFZ data read error");
-
-                std::stringstream ss(str);
-                std::string file;
-                while (std::getline(ss, file, '\0')) // split the strings
-                    mNif.push_back(file);
-
                 break;
             }
             case ESM4::SUB_NIFT:
@@ -149,15 +142,8 @@ void ESM4::Creature::load(ESM4::Reader& reader)
             }
             case ESM4::SUB_KFFZ:
             {
-                std::string str;
+                if (!reader.getZeroTerminatedStringArray(mKf))
-                if (!reader.getZString(str))
                     throw std::runtime_error ("CREA KFFZ data read error");
-
-                std::stringstream ss(str);
-                std::string file;
-                while (std::getline(ss, file, '\0')) // split the strings
-                    mKf.push_back(file);
-
                 break;
             }
             case ESM4::SUB_TPLT: reader.get(mBaseTemplate); break; // FO3

components/esm4/loadnpc.cpp

@@ -144,19 +144,12 @@ void ESM4::Npc::load(ESM4::Reader& reader)
             case ESM4::SUB_MODB: reader.get(mBoundRadius); break;
             case ESM4::SUB_KFFZ:
             {
-                std::string str;
-                if (!reader.getZString(str))
-                    throw std::runtime_error ("NPC_ KFFZ data read error");
-
                 // Seems to be only below 3, and only happens 3 times while loading TES4:
                 //   Forward_SheogorathWithCane.kf
                 //   TurnLeft_SheogorathWithCane.kf
                 //   TurnRight_SheogorathWithCane.kf
-                std::stringstream ss(str);
+                if (!reader.getZeroTerminatedStringArray(mKf))
-                std::string file;
+                    throw std::runtime_error ("NPC_ KFFZ data read error");
-                while (std::getline(ss, file, '\0')) // split the strings
-                    mKf.push_back(file);
-
                 break;
             }
             case ESM4::SUB_LNAM: reader.get(mHairLength); break;

components/esm4/reader.cpp

@@ -693,4 +693,30 @@ bool Reader::getStringImpl(std::string& str, std::size_t size,
     return false; // FIXME: throw instead?
 }
 
+bool Reader::getZeroTerminatedStringArray(std::vector<std::string>& values)
+{
+    const std::size_t size = mCtx.subRecordHeader.dataSize;
+    std::string input(size, '\0');
+    mStream->read(input.data(), size);
+
+    if (mStream->gcount() != static_cast<std::streamsize>(size))
+        return false;
+
+    std::string_view inputView(input.data(), input.size());
+    std::string buffer;
+    while (true)
+    {
+        std::string_view value(inputView.data());
+        const std::size_t next = inputView.find_first_not_of('\0', value.size());
+        if (mEncoder != nullptr)
+            value = mEncoder->getUtf8(value, ToUTF8::BufferAllocationPolicy::UseGrowFactor, buffer);
+        values.emplace_back(value);
+        if (next == std::string_view::npos)
+            break;
+        inputView = inputView.substr(next);
+    }
+
+    return true;
+}
+
 }

components/esm4/reader.hpp

@@ -293,6 +293,8 @@ namespace ESM4 {
             return getStringImpl(str, mCtx.subRecordHeader.dataSize, *mStream, mEncoder);
         }
 
+        bool getZeroTerminatedStringArray(std::vector<std::string>& values);
+
         void enterGroup();
         void exitGroupCheck();