feat(text): Add Unicode paragraph base direction detection

CMakeLists.txt

@@ -79,6 +79,7 @@ add_subdirectory(Dependencies/Precompiled)
 add_subdirectory(Dependencies/Utility)
 add_subdirectory(Dependencies/Bink)
 add_subdirectory(Dependencies/Miles)
+add_subdirectory(Dependencies/UnicodeBidi)
 if (WIN32)
     add_subdirectory(Dependencies/DbgHelp)
     add_subdirectory(Dependencies/Usp10)

Core/Libraries/Source/WWVegas/CMakeLists.txt

@@ -10,6 +10,7 @@ target_link_libraries(core_wwcommon INTERFACE
     d3d8lib
     deps_dbghelp
     deps_miles
+    deps_unicodebidi
     deps_usp10
     stlport
 )

Dependencies/UnicodeBidi/CMakeLists.txt

@@ -0,0 +1,17 @@
+set(UNICODEBIDI_SRC
+    unicodebidi.cpp
+    unicodebidi.h
+    bididirection.inl
+)
+
+add_library(deps_unicodebidi STATIC)
+
+target_sources(deps_unicodebidi PRIVATE ${UNICODEBIDI_SRC})
+
+target_include_directories(deps_unicodebidi PUBLIC
+    ${CMAKE_CURRENT_SOURCE_DIR}
+)
+
+target_link_libraries(deps_unicodebidi PRIVATE
+    deps_precompiled # Gives force included headers. Link early.
+)

Dependencies/UnicodeBidi/README.md

@@ -0,0 +1,32 @@
+# Unicode paragraph direction
+
+`Get_Paragraph_Base_Level` provides the initial paragraph level for complex text
+layout, such as Uniscribe itemization. It implements the first-strong
+decision in [UAX #9 P2/P3](https://www.unicode.org/reports/tr9/tr9-51.html#P2),
+with scanning bounded to the first paragraph. Shaping and visual reordering are
+left to the renderer.
+
+The scan skips nested directional-isolate contents, decodes UTF-16 surrogate
+pairs, and ignores lone surrogates. It also accepts direct supplementary scalar
+values when `wchar_t` is 32 bits. The first L character selects level 0; the first
+R or AL character selects level 1. A paragraph without a strong character
+defaults to level 0.
+
+Both BMP and supplementary classification use the same Unicode 17.0.0 data,
+independent of the operating system. The compact table merges R and AL and
+treats every other non-L class as neutral for this decision. Omitted ranges
+default to L, including the source's `@missing` defaults.
+
+## Data and license
+
+Download [DerivedBidiClass.txt](https://www.unicode.org/Public/17.0.0/ucd/extracted/DerivedBidiClass.txt)
+and regenerate or check the table from the repository root:
+
+```
+python Dependencies/UnicodeBidi/generate_bidi.py DerivedBidiClass.txt
+python Dependencies/UnicodeBidi/generate_bidi.py DerivedBidiClass.txt --check
+```
+
+The generator validates the source version and SHA-256 before extracting ranges.
+The generated data retains Unicode's copyright notice; `unicode-license.txt`
+contains the full Unicode License v3.