Audiolib/backend/app/services/matching/dnb.py

"""
Deutsche Nationalbibliothek (DNB) SRU-Schnittstelle.
Sucht Hörbücher (mat=ton) über MARC21-XML.
"""
import re
import httpx
from xml.etree import ElementTree as ET
from .base import MatchResult

DNB_SRU = "https://services.dnb.de/sru/dnb"
HEADERS = {"User-Agent": "audiolib/1.0 (contact@audiolib.local)"}
_NS_SRW = "http://www.loc.gov/zing/srw/"
_NS_MARC = "http://www.loc.gov/MARC21/slim"


async def search_dnb(title: str, author: str | None = None) -> list[MatchResult]:
    parts = [f'tit="{title}"', "mat=ton"]
    if author:
        parts.append(f'per="{author}"')
    query = " AND ".join(parts)

    params = {
        "version": "1.1",
        "operation": "searchRetrieve",
        "query": query,
        "recordSchema": "MARC21-xml",
        "maximumRecords": "5",
    }
    async with httpx.AsyncClient(headers=HEADERS, timeout=15) as client:
        try:
            r = await client.get(DNB_SRU, params=params)
            r.raise_for_status()
        except Exception:
            return []

    try:
        root = ET.fromstring(r.text)
    except ET.ParseError:
        return []

    results = []
    for record in root.findall(f".//{{{_NS_SRW}}}record"):
        marc = record.find(f".//{{{_NS_MARC}}}record")
        if marc is None:
            continue
        try:
            result = _parse_marc(marc)
            if result:
                results.append(result)
        except Exception:
            continue
    return results


def _field(marc, tag: str, code: str | None = None) -> str | None:
    for f in marc.findall(f"{{{_NS_MARC}}}datafield[@tag='{tag}']"):
        if code:
            sf = f.find(f"{{{_NS_MARC}}}subfield[@code='{code}']")
            if sf is not None and sf.text:
                return sf.text.strip()
        else:
            parts = [sf.text.strip() for sf in f.findall(f"{{{_NS_MARC}}}subfield") if sf.text]
            if parts:
                return " ".join(parts)
    return None


def _fields(marc, tag: str, code: str) -> list[str]:
    out = []
    for f in marc.findall(f"{{{_NS_MARC}}}datafield[@tag='{tag}']"):
        sf = f.find(f"{{{_NS_MARC}}}subfield[@code='{code}']")
        if sf is not None and sf.text:
            out.append(sf.text.strip())
    return out


def _parse_marc(marc) -> MatchResult | None:
    title_a = (_field(marc, "245", "a") or "").rstrip("/ ").strip()
    title_b = _field(marc, "245", "b")
    title = (title_a + " " + title_b.rstrip("/ ").strip()).strip() if title_b else title_a
    if not title:
        return None

    subtitle = title_b.rstrip("/ ").strip() if title_b else None

    author = _field(marc, "100", "a")
    if author:
        author = author.rstrip(",").strip()

    # Sprecher aus 700 $e = "Sprecher" oder $4 = "spk"
    narrator = None
    for f in marc.findall(f"{{{_NS_MARC}}}datafield[@tag='700']"):
        e_sf = f.find(f"{{{_NS_MARC}}}subfield[@code='e']")
        r_sf = f.find(f"{{{_NS_MARC}}}subfield[@code='4']")
        is_narrator = (
            (e_sf is not None and e_sf.text and "prech" in e_sf.text.lower())
            or (r_sf is not None and r_sf.text == "spk")
        )
        if is_narrator:
            n_sf = f.find(f"{{{_NS_MARC}}}subfield[@code='a']")
            if n_sf is not None and n_sf.text:
                narrator = n_sf.text.rstrip(",").strip()
                break

    publisher = (_field(marc, "264", "b") or "").rstrip(",").strip() or None
    year_raw = _field(marc, "264", "c") or _field(marc, "260", "c")
    publish_year = None
    if year_raw:
        m = re.search(r"\d{4}", year_raw)
        if m:
            publish_year = int(m.group())

    description = _field(marc, "520", "a")
    language = _field(marc, "041", "a")
    genres = _fields(marc, "650", "a")[:5]

    series = _field(marc, "830", "a") or _field(marc, "800", "t")
    series_seq = _field(marc, "830", "v") or _field(marc, "800", "v")

    # DNB-ID aus Kontrollfeld 001
    ctrl = marc.find(f"{{{_NS_MARC}}}controlfield[@tag='001']")
    dnb_id = ctrl.text.strip() if ctrl is not None and ctrl.text else None

    # ISBN für Cover
    isbn_raw = _field(marc, "020", "a") or ""
    isbn = re.sub(r"[^0-9X]", "", isbn_raw.split()[0]) if isbn_raw else None
    cover_url = f"https://portal.dnb.de/opac/mvb/cover?isbn={isbn}" if isbn else None

    return MatchResult(
        source="dnb",
        source_id=dnb_id or f"dnb_{title[:30]}",
        title=title,
        subtitle=subtitle,
        author=author,
        narrator=narrator,
        description=description,
        publisher=publisher,
        publish_year=publish_year,
        language=language,
        genres=genres,
        series=series,
        series_sequence=series_seq,
        cover_url=cover_url,
        confidence=0.65,
    )