#정규화 (Normalization)

import re
import unicodedata
from datetime import date, timedelta
from dateutil.relativedelta import relativedelta  # pip install python-dateutil


def normalize_text(s: str) -> str:
    s = unicodedata.normalize("NFKC", s)   # 전각→반각, ㈜→(주), NFD 자모 → 완성형
    s = s.lower()
    return re.sub(r"\s+", " ", s).strip()


# ---- 한글 수사 → 숫자 ("이천오백" → 2500) ----
DIGITS = {"일": 1, "이": 2, "삼": 3, "사": 4, "오": 5, "육": 6, "칠": 7, "팔": 8, "구": 9}
SMALL = {"십": 10, "백": 100, "천": 1000}
LARGE = {"만": 10_000, "억": 100_000_000}


def korean_number(word: str) -> int:
    total, section, digit = 0, 0, 0
    for ch in word:
        if ch.isdigit():
            digit = digit * 10 + int(ch)
        elif ch in DIGITS:
            digit = DIGITS[ch]
        elif ch in SMALL:
            section += (digit or 1) * SMALL[ch]
            digit = 0
        elif ch in LARGE:
            total += ((section + digit) or 1) * LARGE[ch]
            section, digit = 0, 0
    return total + section + digit


def normalize_money(q: str) -> str:
    return re.sub(r"([일이삼사오육칠팔구십백천만억\d]+)\s*원",
                  lambda m: f"{korean_number(m.group(1))}원", q)


# ---- 상대 날짜 → 절대 기간 (기준일은 반드시 인자로) ----
def normalize_dates(q: str, today: date) -> tuple[str, dict]:
    filters: dict = {}
    if m := re.search(r"([일이삼사오육칠팔구십\d]+)\s*개월\s*(전|이내)", q):
        n = korean_number(m.group(1))
        filters = {"from": today - relativedelta(months=n), "to": today}  # "최근 N개월"로 해석
        q = q.replace(m.group(0), "")
    elif "지난달" in q:
        first = today.replace(day=1)
        filters = {"from": first - relativedelta(months=1), "to": first - timedelta(days=1)}
        q = q.replace("지난달", "")
    elif m := re.search(r"작년\s*(\d{1,2})\s*월\s*(이후)?", q):
        start = date(today.year - 1, int(m.group(1)), 1)
        end = today if m.group(2) else start + relativedelta(months=1) - timedelta(days=1)
        filters = {"from": start, "to": end}
        q = q.replace(m.group(0), "")
    elif "지난주" in q:
        this_monday = today - timedelta(days=today.weekday())
        filters = {"from": this_monday - timedelta(days=7), "to": this_monday - timedelta(days=1)}
        q = q.replace("지난주", "")
    return re.sub(r"\s+", " ", q).strip(), filters


today = date(2026, 9, 15)
q = normalize_money(normalize_text("지난달 주문 내역"))
print(normalize_dates(q, today))
# ('주문 내역', {'from': datetime.date(2026, 8, 1), 'to': datetime.date(2026, 8, 31)})
print(normalize_money(normalize_text("십오만원 이상")))   # 150000원 이상
print(normalize_text("１２０ＭＬ　ＡＢＣ㈜ 제품"))         # 120ml abc(주) 제품