AI_portal/contract-check/registry.py

"""Supplier reputation checks against Czech public registers.

Sources (all free, public):
  - ARES (ares.gov.cz) REST   → existence, name, address, legal form, dates,
                                VAT registration (stavZdrojeDph), insolvency
                                flag (stavZdrojeIr), and name→IČO search.
  - MFČR CRPDPH SOAP          → 'nespolehlivý plátce DPH' (unreliable VAT
                                payer) + published bank accounts.
"""
import logging
import re
import xml.etree.ElementTree as ET

import httpx

logger = logging.getLogger(__name__)

ARES_BASE = "https://ares.gov.cz/ekonomicke-subjekty-v-be/rest/ekonomicke-subjekty"
MFCR_URL = ("https://adisrws.mfcr.cz/adistc/axis2/services/"
            "rozhraniCRPDPH.rozhraniCRPDPHSOAP")
_TIMEOUT = 20.0


def _clean_ico(ico: str) -> str:
    return re.sub(r"\D", "", str(ico or ""))


def ares_by_ico(ico: str) -> dict | None:
    """Look up one subject by IČO. Returns a normalized dict or None."""
    ico = _clean_ico(ico).zfill(8) if _clean_ico(ico) else ""
    if len(ico) != 8:
        return None
    try:
        r = httpx.get(f"{ARES_BASE}/{ico}",
                      headers={"accept": "application/json"}, timeout=_TIMEOUT)
        if r.status_code == 404:
            return None
        r.raise_for_status()
        d = r.json()
    except Exception as exc:
        logger.warning("ARES lookup failed for %s: %s", ico, exc)
        return None
    reg = d.get("seznamRegistraci") or {}
    sidlo = (d.get("sidlo") or {}).get("textovaAdresa")
    return {
        "ico": d.get("ico"),
        "dic": d.get("dic"),
        "name": d.get("obchodniJmeno"),
        "address": sidlo,
        "legal_form": d.get("pravniForma"),
        "founded": d.get("datumVzniku"),
        "dissolved": d.get("datumZaniku"),
        # AKTIVNI in the business/commercial register or RES
        "active": reg.get("stavZdrojeVr") == "AKTIVNI"
                  or reg.get("stavZdrojeRes") == "AKTIVNI",
        "vat_registered": reg.get("stavZdrojeDph") == "AKTIVNI",
        "in_insolvency": reg.get("stavZdrojeIr") == "AKTIVNI",
    }


def ares_search_by_name(name: str, limit: int = 5) -> list[dict]:
    """Find candidates by trade name. Returns [{ico, name, address}]."""
    name = (name or "").strip()
    if len(name) < 3:
        return []
    try:
        r = httpx.post(f"{ARES_BASE}/vyhledat",
                       json={"obchodniJmeno": name, "pocet": limit},
                       headers={"accept": "application/json"}, timeout=_TIMEOUT)
        r.raise_for_status()
        d = r.json()
    except Exception as exc:
        logger.warning("ARES name search failed for %r: %s", name, exc)
        return []
    out = []
    for s in (d.get("ekonomickeSubjekty") or [])[:limit]:
        out.append({
            "ico": s.get("ico"),
            "name": s.get("obchodniJmeno"),
            "address": (s.get("sidlo") or {}).get("textovaAdresa"),
        })
    return out


def mfcr_unreliable_vat(ico_or_dic: str) -> dict:
    """MFČR 'nespolehlivý plátce DPH' check. DIČ number = CZ + IČO usually,
    the service takes the numeric part. Returns {checked, unreliable, accounts}."""
    num = _clean_ico(ico_or_dic)
    if len(num) < 8:
        return {"checked": False, "unreliable": None, "accounts": []}
    body = (
        '<?xml version="1.0" encoding="UTF-8"?>'
        '<soapenv:Envelope xmlns:soapenv="http://schemas.xmlsoap.org/soap/envelope/"'
        ' xmlns:urn="http://adis.mfcr.cz/rozhraniCRPDPH/">'
        '<soapenv:Body><urn:StatusNespolehlivyPlatceRequest>'
        f'<urn:dic>{num}</urn:dic>'
        '</urn:StatusNespolehlivyPlatceRequest></soapenv:Body></soapenv:Envelope>'
    )
    try:
        r = httpx.post(MFCR_URL, content=body,
                       headers={"Content-Type": "text/xml; charset=utf-8",
                                "SOAPAction": ""}, timeout=_TIMEOUT)
        r.raise_for_status()
        # strip namespaces for easy find
        xml = re.sub(r'\sxmlns(:\w+)?="[^"]*"', "", r.text)
        xml = re.sub(r"<(/?)\w+:", r"<\1", xml)
        root = ET.fromstring(xml)
        node = root.find(".//statusPlatceDPH")
        if node is None:
            return {"checked": False, "unreliable": None, "accounts": []}
        flag = node.get("nespolehlivyPlatce")  # ANO/NE/NENALEZEN
        accounts = []
        for u in node.findall(".//ucet"):
            std = u.find("standardniUcet")
            nst = u.find("nestandardniUcet")
            if std is not None:
                pref = std.get("predcisli")
                cislo = std.get("cislo")
                kb = std.get("kodBanky")
                acc = (f"{pref}-" if pref else "") + f"{cislo}/{kb}"
                accounts.append(acc)
            elif nst is not None:
                accounts.append(nst.get("cislo"))
        return {
            "checked": flag in ("ANO", "NE"),
            "unreliable": (flag == "ANO") if flag in ("ANO", "NE") else None,
            "vat_status": flag,
            "accounts": accounts,
        }
    except Exception as exc:
        logger.warning("MFČR VAT check failed for %s: %s", num, exc)
        return {"checked": False, "unreliable": None, "accounts": []}


def reputation(ico: str) -> dict | None:
    """Full reputation report for one IČO. None if the subject doesn't exist."""
    ares = ares_by_ico(ico)
    if ares is None:
        return None
    vat = mfcr_unreliable_vat(ares.get("dic") or ico)
    # Simple flag set for the UI to highlight.
    flags = []
    if ares.get("dissolved"):
        flags.append(("danger", "Subjekt zanikl"))
    if not ares.get("active"):
        flags.append(("danger", "Není aktivní v rejstříku"))
    if ares.get("in_insolvency"):
        flags.append(("danger", "Vedeno v insolvenčním rejstříku"))
    if vat.get("unreliable") is True:
        flags.append(("danger", "Nespolehlivý plátce DPH"))
    if not flags:
        flags.append(("ok", "Bez zjištěných rizik v rejstřících"))
    return {"ares": ares, "vat": vat, "flags": flags}