EpiSupport-Zammad-Scraper/app/conversation.py

"""Konversation (Artikel) eines Tickets: in der DB zwischenspeichern und laden.

So muss die Leseseite nicht live auf Zammad warten (vermeidet Proxy-Timeouts).
Der Poller füllt/aktualisiert die Artikel im Hintergrund; die Webseite liest
nur aus der DB (mit einmaligem Lazy-Load als Rückfall).
"""
import json
import logging
from datetime import datetime, timezone

from . import config, db, textutil

log = logging.getLogger("episupport.conversation")


def _now() -> str:
    return datetime.now(timezone.utc).isoformat()


ATT_CACHE = config.DATA_DIR / "att_cache"
ATT_CACHE.mkdir(exist_ok=True)


def _attachment_meta(article) -> list[dict]:
    """Relevante Anhang-Metadaten (id, filename, mime) je Artikel."""
    out = []
    for x in (article.get("attachments") or []):
        fn = x.get("filename")
        if not fn or fn == "message.html":
            continue
        prefs = x.get("preferences") or {}
        mime = prefs.get("Mime-Type") or prefs.get("Content-Type") or ""
        out.append({"id": x.get("id"), "filename": fn, "mime": mime})
    return out


def save_articles(ticket_id: int, articles: list[dict], conn=None) -> int:
    """Bereits abgerufene Artikel in die DB schreiben (INSERT OR REPLACE).

    Wird `conn` übergeben, wird diese bestehende Verbindung genutzt (wichtig
    im Poller, der bereits eine Transaktion offen hält – sonst „database is
    locked"). Ohne `conn` wird eine eigene, kurze Transaktion verwendet.
    """
    now = _now()

    def _write(c):
        for a in articles:
            c.execute(
                "INSERT OR REPLACE INTO articles (article_id, ticket_id, sender, "
                "frm, subject, body, content_type, internal, created_at, "
                "attachments, fetched_at) VALUES (?,?,?,?,?,?,?,?,?,?,?)",
                (a.get("id"), ticket_id, a.get("sender"), a.get("from"),
                 a.get("subject"), a.get("body"), a.get("content_type"),
                 1 if a.get("internal") else 0, a.get("created_at"),
                 json.dumps(_attachment_meta(a)), now),
            )

    if conn is not None:
        _write(conn)
    else:
        with db.get_conn() as own:
            _write(own)
    return len(articles)


def store_articles(client, ticket_id: int, timeout: int | None = None,
                   conn=None) -> int:
    """Artikel live von Zammad holen und speichern."""
    articles = client.ticket_articles(ticket_id, timeout=timeout)
    return save_articles(ticket_id, articles, conn=conn)


def load_articles(ticket_id: int) -> list[dict]:
    """Aufbereitete Artikel aus der DB (HTML sicher zu Text bereinigt)."""
    with db.get_conn() as conn:
        rows = conn.execute(
            "SELECT * FROM articles WHERE ticket_id=? ORDER BY created_at, article_id",
            (ticket_id,)
        ).fetchall()
    out = []
    for r in rows:
        body = r["body"] or ""
        if (r["content_type"] or "").startswith("text/html"):
            body = textutil.clean_html(body)
        out.append({
            "id": r["article_id"],
            "sender": r["sender"],
            "frm": r["frm"] or "",
            "subject": r["subject"],
            "created_at": r["created_at"],
            "internal": r["internal"],
            "body": body,
            "attachments": _parse_attachments(r["attachments"]),
        })
    return out


def _parse_attachments(raw) -> list[dict]:
    """Verträgt altes Format (nur Dateinamen) und neues (id/filename/mime)."""
    items = json.loads(raw or "[]")
    out = []
    for x in items:
        if isinstance(x, str):  # altes Cache-Format
            out.append({"id": None, "filename": x, "mime": "", "is_image": False})
        else:
            mime = x.get("mime") or ""
            out.append({
                "id": x.get("id"),
                "filename": x.get("filename"),
                "mime": mime,
                "is_image": mime.startswith("image/"),
            })
    return out


def find_attachment(article_id: int, att_id: int) -> dict | None:
    """Metadaten eines Anhangs (Mime/Dateiname) aus der DB – für die Auslieferung."""
    with db.get_conn() as conn:
        row = conn.execute(
            "SELECT attachments FROM articles WHERE article_id=?", (article_id,)
        ).fetchone()
    if not row:
        return None
    for a in _parse_attachments(row["attachments"]):
        if a["id"] == att_id:
            return a
    return None


def get_attachment_bytes(client, ticket_id: int, article_id: int, att_id: int) -> bytes:
    """Anhang-Bytes liefern; einmal von Zammad holen und auf Platte cachen."""
    cache = ATT_CACHE / f"{article_id}_{att_id}"
    if cache.exists():
        return cache.read_bytes()
    data = client.get_attachment(ticket_id, article_id, att_id)
    try:
        cache.write_bytes(data)
    except OSError:
        pass
    return data


def has_articles(ticket_id: int) -> bool:
    with db.get_conn() as conn:
        return conn.execute(
            "SELECT 1 FROM articles WHERE ticket_id=? LIMIT 1", (ticket_id,)
        ).fetchone() is not None


def tickets_without_articles(limit: int = 10) -> list[int]:
    with db.get_conn() as conn:
        rows = conn.execute(
            "SELECT id FROM tickets WHERE id NOT IN "
            "(SELECT DISTINCT ticket_id FROM articles) "
            "ORDER BY zammad_updated_at DESC LIMIT ?", (limit,)
        ).fetchall()
    return [r["id"] for r in rows]