Repository navigation
from_utf8() and scan_text() now enforce well-formed UTF-8 per Unicode Table 3-7 (overlong encodings, surrogates and code points above U+10FFFF are rejected; one U+FFFD per maximal subpart). New: from_utf8_left_byte_unconsumed() helper and LIBUNICODE_FROM_UTF8_STRICT feature macro. Callers must re-feed a byte that ended an ill-formed sequence when the helper says it was not consumed. (#160)