The Carmodoo HTML response sometimes contains invalid EUC-KR byte
sequences (e.g., 0xA4 followed by ASCII 'F'). This caused the decoder
to fall back to Latin-1, corrupting all Korean text.
Fixed by using errors='replace' which preserves Korean text while
replacing only the invalid byte sequences with replacement characters.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
209c63e463