"""Pure normalization rules: company-name spellings and invoice-number formats."""
import re


def company_key(name):
    """Grouping key for company-name spellings: case/punctuation/whitespace-insensitive, the German
    legal-form spellings unified (GmbH == Ges.m.b.H. == GesmbH, 'und' == '&'). Different legal forms
    (GmbH vs AG) stay different — near-misses beyond this are only SUGGESTED for manual merge."""
    n = (name or "").strip().lower()
    n = n.replace(" und ", " & ").replace(" u. ", " & ")
    n = re.sub(r"ges\.?\s*m\.?\s*b\.?\s*h\.?", "gmbh", n)
    n = re.sub(r"[^a-z0-9äöüß&]+", "", n)
    return n.replace("gbmh", "gmbh")   # 'GbmH' is a transposition typo, not a different legal form


def invoice_seq(number, year):
    """Canonical integer sequence of an invoice number for the given 4-digit year, whatever the format:
        '202262' -> 62, '2022-62' -> 62, '22-62' -> 62, '62-2015' -> 62 (SEQ-YEAR), '2026-01' -> 1, '62' -> 62.
    Combined / concatenated numbers ('41/46', '5556') return None so they never pass for a single invoice."""
    toks = [t for t in re.split(r"\D+", str(number or "")) if t]
    if not toks:
        return None
    if len(toks) == 1:                                               # bare '25' / glued 'YYYYNN' / 'YYNN'
        s = toks[0]
        if s == year:
            return None                                              # a bare 4-digit year is not a sequence
        if year and s.startswith(year) and len(s) > len(year):
            s = s[len(year):]                                        # 'YYYYNN'
        elif year and s.startswith(year[2:]) and len(s) > 2:
            s = s[2:]                                                # 'YYNN'
        # else keep s verbatim: a bare sequence ('25','62') even when it equals the 2-digit year (invoice #25 in 2025)
    else:                                                            # PREFIX-SEQ / SEQ-YEAR: drop ONE year token
        non_year = list(toks)
        for yt in (year, year[2:]):
            if yt in non_year:
                non_year.remove(yt)
                break
        if len(non_year) != 1:
            return None                                              # combined ('41','46') -> no single sequence
        s = non_year[0]
    s = s.lstrip("0") or s
    return int(s) if s.isdigit() and len(s) <= 3 else None


def invoice_year(number):
    """The 4-digit year an invoice number encodes: '2025-46'->'2025', '25-46'/'26-01'->20YY, else None. A
    New-Year invoice dated in January can still carry the old year's number; this is that year."""
    s = str(number or "")
    m = re.search(r"(?:19|20)\d\d", s)
    if m:
        return m.group(0)
    m = re.match(r"\s*(\d{2})\D", s)                      # 'YY-NN' 2-digit prefix
    return "20" + m.group(1) if m else None