#!/usr/bin/env python3
"""CASAN H4 — Unicode confusable / obfuscation normalizer (Track A, V3).

Reads text on stdin, writes a normalized variant on stdout used ONLY for
injection/jailbreak phrase matching (never for PII/secret redaction, so we
never widen what counts as a secret).

Deterministic transforms, in order:
  1. NFKC compatibility normalization  — folds fullwidth (ｉｇｎｏｒｅ) and
     other compatibility forms to their ASCII equivalents.
  2. Strip zero-width / BOM / soft-hyphen formatting characters that split a
     word so a blocklist never sees it (i·g·n·o·r·e).
  3. Fold a fixed table of common Cyrillic/Greek homoglyphs to their Latin
     lookalikes (іgnоrе -> ignore). NFKC does NOT do this — confusables are a
     separate Unicode concern — so the table is explicit and auditable.

The table is intentionally small and covers the lookalikes actually used in
prompt-injection homoglyph attacks; extend it as new vectors appear.
"""
import sys
import unicodedata

# Zero-width, BOM, and invisible formatting code points.
ZERO_WIDTH = {
    0x200B,  # zero-width space
    0x200C,  # zero-width non-joiner
    0x200D,  # zero-width joiner
    0x2060,  # word joiner
    0xFEFF,  # BOM / zero-width no-break space
    0x00AD,  # soft hyphen
    0x180E,  # Mongolian vowel separator
    0x2061, 0x2062, 0x2063, 0x2064,  # invisible math operators
}

# Common Cyrillic / Greek homoglyphs -> Latin lookalike. Lowercase and
# uppercase both listed because matching is case-insensitive downstream but the
# fold must run before case handling to be safe.
CONFUSABLES = {
    # Cyrillic lowercase
    "а": "a", "е": "e", "о": "o", "р": "p", "с": "c", "у": "y", "х": "x",
    "ѕ": "s", "і": "i", "ј": "j", "ԁ": "d", "һ": "h", "ӏ": "l", "п": "n",
    "г": "r", "т": "t", "к": "k", "м": "m", "в": "b",
    # Cyrillic uppercase
    "А": "A", "В": "B", "Е": "E", "К": "K", "М": "M", "Н": "H", "О": "O",
    "Р": "P", "С": "C", "Т": "T", "У": "Y", "Х": "X", "Ѕ": "S", "І": "I",
    "Ј": "J",
    # Greek
    "α": "a", "ε": "e", "ο": "o", "ρ": "p", "τ": "t", "ν": "v", "κ": "k",
    "Α": "A", "Β": "B", "Ε": "E", "Ζ": "Z", "Η": "H", "Ι": "I", "Κ": "K",
    "Μ": "M", "Ν": "N", "Ο": "O", "Ρ": "P", "Τ": "T", "Υ": "Y", "Χ": "X",
}


def normalize(text: str) -> str:
    text = unicodedata.normalize("NFKC", text)
    text = "".join(ch for ch in text if ord(ch) not in ZERO_WIDTH)
    text = "".join(CONFUSABLES.get(ch, ch) for ch in text)
    return text


if __name__ == "__main__":
    sys.stdout.write(normalize(sys.stdin.read()))