Tell MSVC to compile NVDAHelper sources as UTF-8 (#21032)
Fixes #20584.
Summary of the issue:
MSVC can read UTF-8 source files without a byte-order mark using the current Windows code page. On systems that use a different code page, it can misread characters in cppjieba headers and report C++ syntax errors.
Description of user facing changes:
No intended changes to NVDA’s interface or commands.
Description of developer facing changes:
The NVDAHelper build explicitly uses UTF-8 instead of relying on the Windows code page. This makes the compiler read UTF-8 source files consistently across build machines.
Description of development approach:
Added /utf-8 to the common CCFLAGS in nvdaHelper/archBuild_sconscript. The third-party build environment inherits these flags, so the option also applies to cppjieba.
MSVC’s /utf-8 option sets both the source and execution character sets. It therefore also determines the encoding of ordinary string literals in compiled code.
