nx_utf8.nx

utf8.nx -- UTF-8 validator + codepoint decoder (RFC 3629). Needed by: X.509 UTF8String fields, HTTP headers, JSON strings, NishiLang source code itself. Without a validator we either accept malformed sequences (security problem: overlong forms can hide dangerous characters) or reject valid input (bugs). UTF-8 byte patterns (RFC 3629 §3): 0xxxxxxx 1 byte, U+0000..U+007F 110xxxxx 10xxxxxx 2 bytes, U+0080..U+07FF 1110xxxx 10xxxxxx 10xxxxxx 3 bytes, U+0800..U+FFFF 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx 4 bytes, U+10000..U+10FFFF RFC 3629 §4 additional constraints: - Reject overlong encodings (fewer bytes than minimum for the codepoint). E.g., U+0000 must be `00`, not `C0 80`. - Reject surrogates U+D800..U+DFFF (reserved for UTF-16). - Reject codepoints > U+10FFFF. Invariants: U1 utf8_validate returns exact byte position of first invalid byte, or input length on clean success. U2 utf8_decode_one advances the caller's position by the number of bytes consumed; returns negative codepoint on invalid sequence. U3 No reads beyond the input's declared length; no buffer overruns possible from malformed input.