Fix banner text formatting, import / export

src/openrct2/core/String.cpp

@@ -509,6 +509,13 @@ namespace String
         return utf8_write_codepoint(dst, codepoint);
     }
 
+    void AppendCodepoint(std::string& str, codepoint_t codepoint)
+    {
+        char buffer[8]{};
+        utf8_write_codepoint(buffer, codepoint);
+        str.append(buffer);
+    }
+
     bool IsWhiteSpace(codepoint_t codepoint)
     {
         // 0x3000 is the 'ideographic space', a 'fullwidth' character used in CJK languages.

src/openrct2/core/String.hpp

@@ -94,6 +94,7 @@ namespace String
     codepoint_t GetNextCodepoint(utf8* ptr, utf8** nextPtr = nullptr);
     codepoint_t GetNextCodepoint(const utf8* ptr, const utf8** nextPtr = nullptr);
     utf8* WriteCodepoint(utf8* dst, codepoint_t codepoint);
+    void AppendCodepoint(std::string& str, codepoint_t codepoint);
 
     bool IsWhiteSpace(codepoint_t codepoint);
     utf8* Trim(utf8* str);

src/openrct2/rct1/S4Importer.cpp

@@ -3006,7 +3006,8 @@ private:
         const auto originalString = _s4.string_table[(stringId - USER_STRING_START) % 1024];
         std::string_view originalStringView(originalString, USER_STRING_MAX_LENGTH);
         auto asUtf8 = rct2_to_utf8(originalStringView, RCT2_LANGUAGE_ID_ENGLISH_UK);
-        return asUtf8.data();
+        auto justText = RCT12RemoveFormattingUTF8(asUtf8);
+        return justText.data();
     }
 
     void FixLandOwnership()

src/openrct2/rct12/RCT12.cpp

@@ -9,6 +9,7 @@
 
 #include "RCT12.h"
 
+#include "../core/String.hpp"
 #include "../localisation/Localisation.h"
 #include "../ride/Track.h"
 #include "../world/Banner.h"
@@ -1041,6 +1042,24 @@ RCT12RideId OpenRCT2RideIdToRCT12RideId(const ride_id_t rideId)
     return rideId;
 }
 
+static bool RCT12IsFormatChar(codepoint_t c)
+{
+    if (c >= RCT2_STRING_FORMAT_ARG_START && c <= RCT2_STRING_FORMAT_ARG_END)
+    {
+        return true;
+    }
+    if (c >= RCT2_STRING_FORMAT_COLOUR_START && c <= RCT2_STRING_FORMAT_COLOUR_END)
+    {
+        return true;
+    }
+    return false;
+}
+
+static bool RCT12IsFormatChar(char c)
+{
+    return RCT12IsFormatChar(static_cast<codepoint_t>(c));
+}
+
 bool IsLikelyUTF8(std::string_view s)
 {
     // RCT2 uses CP-1252 so some characters may be >= 128. However we don't expect any
@@ -1048,15 +1067,28 @@ bool IsLikelyUTF8(std::string_view s)
     // that the string is UTF-8.
     for (auto c : s)
     {
-        auto chr = static_cast<uint8_t>(c);
-        if (chr >= RCT2_STRING_FORMAT_ARG_START && chr <= RCT2_STRING_FORMAT_ARG_END)
-        {
-            return true;
-        }
-        if (chr >= RCT2_STRING_FORMAT_COLOUR_START && chr <= RCT2_STRING_FORMAT_COLOUR_END)
+        if (RCT12IsFormatChar(c))
         {
             return true;
         }
     }
     return false;
 }
+
+std::string RCT12RemoveFormattingUTF8(std::string_view s)
+{
+    std::string result;
+    result.reserve(s.size() * 2);
+
+    CodepointView codepoints(s);
+    for (auto codepoint : codepoints)
+    {
+        if (!RCT12IsFormatChar(codepoint))
+        {
+            String::AppendCodepoint(result, codepoint);
+        }
+    }
+
+    result.shrink_to_fit();
+    return result;
+}

src/openrct2/rct12/RCT12.h

@@ -868,3 +868,4 @@ RCT12ObjectEntryIndex OpenRCT2EntryIndexToRCTEntryIndex(const ObjectEntryIndex i
 ride_id_t RCT12RideIdToOpenRCT2RideId(const RCT12RideId rideId);
 RCT12RideId OpenRCT2RideIdToRCT12RideId(const ride_id_t rideId);
 bool IsLikelyUTF8(std::string_view s);
+std::string RCT12RemoveFormattingUTF8(std::string_view s);

src/openrct2/rct2/S6Exporter.cpp

@@ -1367,12 +1367,13 @@ void S6Exporter::ExportBanner(RCT12Banner& dst, const Banner& src)
 
         dst.string_idx = STR_DEFAULT_SIGN;
 
-        auto bannerText = src.text;
+        std::string bannerText;
         if (!(src.flags & BANNER_FLAG_IS_WALL) && !(src.flags & BANNER_FLAG_IS_LARGE_SCENERY))
         {
-            auto formatCode = static_cast<char>(RCT2_STRING_FORMAT_COLOUR_START + src.text_colour);
-            bannerText.insert(bannerText.begin(), formatCode);
+            auto formatCode = static_cast<codepoint_t>(RCT2_STRING_FORMAT_COLOUR_START + src.text_colour);
+            String::AppendCodepoint(bannerText, formatCode);
         }
+        bannerText.append(src.text);
 
         auto stringId = AllocateUserString(bannerText);
         if (stringId != std::nullopt)

src/openrct2/rct2/S6Importer.cpp

@@ -1661,7 +1661,8 @@ public:
         const auto originalString = _s6.custom_strings[(stringId - USER_STRING_START) % 1024];
         std::string_view originalStringView(originalString, USER_STRING_MAX_LENGTH);
         auto asUtf8 = rct2_to_utf8(originalStringView, RCT2_LANGUAGE_ID_ENGLISH_UK);
-        return asUtf8.data();
+        auto justText = RCT12RemoveFormattingUTF8(asUtf8);
+        return justText.data();
     }
 
     std::vector<rct_object_entry> GetRequiredObjects()

src/openrct2/world/Banner.cpp

@@ -46,8 +46,15 @@ void Banner::FormatTextTo(Formatter& ft, bool addColour) const
     {
         auto formatToken = FormatTokenFromTextColour(text_colour);
         auto tokenText = FormatTokenToString(formatToken);
+
+        thread_local std::string tokenTextColourBuffer;
+        tokenTextColourBuffer.clear();
+        tokenTextColourBuffer.push_back('{');
+        tokenTextColourBuffer.append(tokenText);
+        tokenTextColourBuffer.push_back('}');
+
         ft.Add<rct_string_id>(STR_STRING_STRINGID);
-        ft.Add<const char*>(tokenText.data());
+        ft.Add<const char*>(tokenTextColourBuffer.c_str());
     }
 
     FormatTextTo(ft);