business-card-scanner/app/main.py

"""HTTP-Schnittstelle und Auslieferung der PWA."""
import asyncio
import logging
import uuid
from contextlib import asynccontextmanager
from datetime import datetime, timezone
from pathlib import Path

from fastapi import Depends, FastAPI, Form, HTTPException, Request, Response, UploadFile
from fastapi.responses import FileResponse, JSONResponse, PlainTextResponse
from fastapi.staticfiles import StaticFiles

from . import (
    config, db, duplicates, extract, geocode, imaging, reminders, security,
    segment, vcard,
)

logging.basicConfig(level=logging.INFO, format="%(asctime)s %(levelname)s %(name)s %(message)s")
log = logging.getLogger("bcs")

STATIC_DIR = Path(__file__).resolve().parent.parent / "static"
MAX_UPLOAD_BYTES = 30 * 1024 * 1024

EDITABLE = [
    "first_name", "last_name", "job_title", "company", "department",
    "email", "phone", "mobile", "fax", "website",
    "street", "postal_code", "city", "country", "notes", "occasion", "place",
]


@asynccontextmanager
async def lifespan(app: FastAPI):
    config.ensure_dirs()
    db.connect()
    if not security.enabled():
        log.warning("APP_PASSWORD ist leer - die App ist ohne Anmeldung erreichbar.")
    if not extract.available():
        log.warning("ANTHROPIC_API_KEY fehlt - Karten werden nur zugeschnitten, nicht ausgelesen.")

    # Karten, die beim letzten Herunterfahren noch in der Warteschlange
    # standen, wuerden sonst dauerhaft auf "wird gelesen" stehen bleiben.
    unfinished = db.query("SELECT id FROM cards WHERE extract_status = 'offen'")
    if unfinished:
        log.info("%d Karte(n) aus dem letzten Lauf werden nachgeholt", len(unfinished))
        start_reading([row["id"] for row in unfinished])

    task = asyncio.create_task(reminders.run_forever())
    try:
        yield
    finally:
        task.cancel()


app = FastAPI(title="Business Card Scanner", lifespan=lifespan, docs_url=None, redoc_url=None)
auth = Depends(security.require_auth)


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


def card_dict(row) -> dict:
    card = dict(row)
    card["image_url"] = f"/api/cards/{card['id']}/image" if card.get("image_file") else None
    card["back_image_url"] = (
        f"/api/cards/{card['id']}/image?side=back" if card.get("back_image_file") else None
    )
    card["has_contact"] = any(
        card.get(f) for f in ("last_name", "company", "email", "phone", "mobile")
    )
    return card


# --------------------------------------------------------------------------- Sitzung

@app.get("/healthz")
def healthz() -> dict:
    db.query_one("SELECT 1 AS ok")
    return {"status": "ok"}


@app.get("/api/session")
def session(request: Request) -> dict:
    return {
        "auth_required": security.enabled(),
        "authenticated": not security.enabled()
        or security.valid_token(request.cookies.get(security.COOKIE)),
        "extraction_available": extract.available(),
        "mail_configured": bool(config.SMTP_HOST and config.REMINDER_TO),
    }


@app.post("/api/login")
def login(response: Response, password: str = Form(...)) -> dict:
    if not security.enabled():
        return {"ok": True}
    if not security.check_password(password):
        raise HTTPException(status_code=401, detail="Passwort stimmt nicht")
    response.set_cookie(
        security.COOKIE,
        security.issue_token(),
        max_age=config.SESSION_DAYS * 86400,
        httponly=True,
        samesite="lax",
        secure=True,
        path="/",
    )
    return {"ok": True}


@app.post("/api/logout")
def logout(response: Response) -> dict:
    response.delete_cookie(security.COOKIE, path="/")
    return {"ok": True}


# ------------------------------------------------------------------------------ Scan

def _process_photo(raw: bytes) -> tuple:
    """Rechenlastiger Teil: dekodieren, freistellen, als JPEG kodieren.

    Die Zuschnitte werden einzeln kodiert und sofort freigegeben - bei zwanzig
    Karten liegen sonst rund 70 MB Rohbild gleichzeitig im Speicher.
    """
    meta = imaging.read_metadata(raw)
    crops, fallback = segment.segment(imaging.decode(raw))
    encoded = []
    while crops:
        encoded.append(imaging.encode_jpeg(crops.pop(0)))
    return meta, encoded, fallback


# Laufende Lesevorgaenge festhalten, sonst raeumt die Garbage Collection sie
# mitten im Lauf weg.
_reading: set = set()


def start_reading(card_ids: list, side: str = "front") -> None:
    task = asyncio.create_task(_read_cards(card_ids, side))
    _reading.add(task)
    task.add_done_callback(_reading.discard)


async def _read_cards(card_ids: list, side: str = "front") -> None:
    """Das Auslesen laeuft nach der Antwort weiter - der Nutzer sieht seine
    Karten sofort, die Felder tropfen nach."""
    column = "back_image_file" if side == "back" else "image_file"
    rows = [db.query_one("SELECT * FROM cards WHERE id = ?", (i,)) for i in card_ids]
    rows = [row for row in rows if row is not None and row[column]]
    images = [(config.IMAGE_DIR / row[column]).read_bytes() for row in rows]

    results = await extract.extract_all(images)

    for row, result in zip(rows, results):
        rotation = int(result.get("rotation") or 0) % 360
        if rotation:
            # Die Lage der Karte auf dem Tisch ist geometrisch nicht
            # bestimmbar - das Modell sagt uns, wie herum sie gehoert.
            path = config.IMAGE_DIR / row[column]
            path.write_bytes(
                await asyncio.to_thread(imaging.rotate_jpeg, path.read_bytes(), rotation)
            )
        fields = {f: (result.get(f) or None) for f in extract.FIELDS}
        # COALESCE, weil das Auslesen laeuft, waehrend die Karte schon
        # sichtbar ist: was in der Zwischenzeit von Hand eingetragen wurde,
        # darf das Modell nicht ueberschreiben.
        assignments = ", ".join(f"{name} = COALESCE({name}, ?)" for name in fields)
        db.execute(
            f"UPDATE cards SET {assignments}, extract_status = ?, extract_error = ? "
            "WHERE id = ? AND extract_status = 'offen'",
            [
                *fields.values(),
                "fehler" if result.get("_error") else "ok",
                result.get("_error"),
                row["id"],
            ],
        )
    log.info("%d Karte(n) ausgelesen", len(rows))


@app.post("/api/scan", dependencies=[auth])
async def scan(
    file: UploadFile,
    lat: float = Form(None),
    lon: float = Form(None),
    occasion: str = Form(""),
) -> JSONResponse:
    raw = await file.read()
    if not raw:
        raise HTTPException(status_code=400, detail="Leere Datei")
    if len(raw) > MAX_UPLOAD_BYTES:
        raise HTTPException(status_code=413, detail="Foto ist groesser als 30 MB")

    try:
        meta, encoded, fallback = await asyncio.to_thread(_process_photo, raw)
    except Exception as exc:
        log.exception("Foto konnte nicht verarbeitet werden")
        raise HTTPException(status_code=400, detail=f"Bild nicht lesbar: {exc}") from exc

    # Der Ort des Fotos schlaegt den des Uploads: Karten werden oft erst
    # abends am Schreibtisch abfotografiert, nicht auf der Messe.
    if meta["lat"] is not None:
        geo_source = "exif"
    elif lat is not None and lon is not None:
        meta["lat"], meta["lon"], geo_source = lat, lon, "browser"
    else:
        geo_source = None
    place = await asyncio.to_thread(geocode.reverse, meta["lat"], meta["lon"])

    batch_id = uuid.uuid4().hex
    created = now_iso()
    source = "einzel" if fallback else "stapel"
    cards = []
    for upright in encoded:
        card_id = uuid.uuid4().hex
        image_file = f"{card_id}.jpg"
        (config.IMAGE_DIR / image_file).write_bytes(upright)
        columns = {
            "id": card_id,
            "created_at": created,
            "captured_at": meta["captured_at"] or created,
            "batch_id": batch_id,
            "image_file": image_file,
            "source": source,
            "notes": "",
            "lat": meta["lat"],
            "lon": meta["lon"],
            "place": place,
            "geo_source": geo_source,
            "occasion": occasion.strip() or None,
            "extract_status": "offen",
        }
        names = ", ".join(columns)
        placeholders = ", ".join("?" for _ in columns)
        db.execute(
            f"INSERT INTO cards ({names}) VALUES ({placeholders})", list(columns.values())
        )
        cards.append(card_dict(db.query_one("SELECT * FROM cards WHERE id = ?", (card_id,))))

    # Antwort geht sofort raus, das Auslesen laeuft danach weiter.
    start_reading([card["id"] for card in cards])

    log.info("Scan %s: %d Karte(n), Fallback=%s", batch_id, len(cards), fallback)
    return JSONResponse(
        {"batch_id": batch_id, "count": len(cards), "fallback": fallback, "cards": cards}
    )


# ----------------------------------------------------------------------------- Karten

@app.get("/api/cards", dependencies=[auth])
def list_cards(q: str = "", filter: str = "alle") -> dict:
    sql = "SELECT * FROM cards"
    where, params = [], []
    if q.strip():
        needle = f"%{q.strip()}%"
        searchable = [
            "first_name", "last_name", "company", "job_title", "email", "phone", "mobile",
            "website", "city", "place", "notes", "department", "street", "occasion",
        ]
        where.append("(" + " OR ".join(f"{c} LIKE ?" for c in searchable) + ")")
        params += [needle] * len(searchable)
    if filter == "offen":
        where.append("exported_at IS NULL")
    elif filter == "unvollstaendig":
        where.append(
            "(extract_status = 'fehler' "
            "OR (extract_status = 'ok' AND last_name IS NULL AND company IS NULL))"
        )
    elif filter == "erinnerung":
        where.append(
            "id IN (SELECT card_id FROM reminders WHERE sent_at IS NULL)"
        )
    if where:
        sql += " WHERE " + " AND ".join(where)
    sql += " ORDER BY COALESCE(captured_at, created_at) DESC, rowid DESC"

    rows = [card_dict(r) for r in db.query(sql, params)]
    totals = db.query_one(
        """
        SELECT COUNT(*) AS total,
               SUM(CASE WHEN exported_at IS NULL THEN 1 ELSE 0 END) AS offen,
               SUM(CASE WHEN extract_status = 'offen' THEN 1 ELSE 0 END) AS laeuft
        FROM cards
        """
    )
    pending = db.query_one(
        "SELECT COUNT(*) AS n FROM reminders WHERE sent_at IS NULL"
    )
    return {
        "cards": rows,
        "stats": {
            "total": totals["total"] or 0,
            "offen": totals["offen"] or 0,
            "erinnerungen": pending["n"] or 0,
            "laeuft": totals["laeuft"] or 0,
        },
    }


def _card_or_404(card_id: str):
    row = db.query_one("SELECT * FROM cards WHERE id = ?", (card_id,))
    if row is None:
        raise HTTPException(status_code=404, detail="Karte nicht gefunden")
    return row


@app.get("/api/cards/{card_id}", dependencies=[auth])
def get_card(card_id: str) -> dict:
    card = card_dict(_card_or_404(card_id))
    card["reminders"] = [
        dict(r)
        for r in db.query(
            "SELECT * FROM reminders WHERE card_id = ? ORDER BY due_at", (card_id,)
        )
    ]
    card["duplicates"] = duplicates.find(card)
    return card


@app.patch("/api/cards/{card_id}", dependencies=[auth])
async def update_card(card_id: str, request: Request) -> dict:
    _card_or_404(card_id)
    payload = await request.json()
    changes = {k: (payload[k] or None) for k in EDITABLE if k in payload}
    if "notes" in changes:
        changes["notes"] = changes["notes"] or ""
    if not changes:
        raise HTTPException(status_code=400, detail="Keine bekannten Felder im Aufruf")
    assignments = ", ".join(f"{k} = ?" for k in changes)
    db.execute(
        f"UPDATE cards SET {assignments} WHERE id = ?", [*changes.values(), card_id]
    )
    return card_dict(_card_or_404(card_id))


@app.delete("/api/cards/{card_id}", dependencies=[auth])
def delete_card(card_id: str) -> dict:
    row = _card_or_404(card_id)
    for name in (row["image_file"], row["back_image_file"]):
        if name:
            (config.IMAGE_DIR / name).unlink(missing_ok=True)
            _original_of(name).unlink(missing_ok=True)
    db.execute("DELETE FROM cards WHERE id = ?", (card_id,))
    return {"ok": True}


@app.get("/api/cards/{card_id}/image", dependencies=[auth])
def card_image(card_id: str, side: str = "front") -> FileResponse:
    row = _card_or_404(card_id)
    name = row["back_image_file"] if side == "back" else row["image_file"]
    path = config.IMAGE_DIR / (name or "")
    if not name or not path.exists():
        raise HTTPException(status_code=404, detail="Kein Bild gespeichert")
    return FileResponse(path, media_type="image/jpeg")


def _original_of(name: str) -> Path:
    """Unbearbeitete Fassung neben dem Bild. Damit bleibt jede Korrektur
    zuruecknehmbar und wird nicht auf eine schon bearbeitete Datei gerechnet."""
    return config.IMAGE_DIR / (name[:-4] + "-orig.jpg")


@app.post("/api/cards/{card_id}/edit", dependencies=[auth])
async def edit_image(card_id: str, request: Request) -> dict:
    row = _card_or_404(card_id)
    payload = await request.json()
    side = payload.get("side", "front")
    name = row["back_image_file"] if side == "back" else row["image_file"]
    if not name:
        raise HTTPException(status_code=404, detail="Kein Bild gespeichert")

    current = config.IMAGE_DIR / name
    original = _original_of(name)
    if not original.exists():
        original.write_bytes(current.read_bytes())

    if payload.get("reset"):
        current.write_bytes(original.read_bytes())
        return card_dict(_card_or_404(card_id))

    rotate = int(payload.get("rotate") or 0)
    if rotate % 90:
        raise HTTPException(status_code=400, detail="Drehung nur in 90-Grad-Schritten")
    angle = max(-20.0, min(20.0, float(payload.get("angle") or 0)))

    crop = payload.get("crop")
    if crop is not None:
        try:
            crop = {key: float(crop[key]) for key in ("x", "y", "w", "h")}
        except (KeyError, TypeError, ValueError) as exc:
            raise HTTPException(status_code=400, detail="Zuschnitt unvollstaendig") from exc
        if crop["w"] <= 0.02 or crop["h"] <= 0.02:
            raise HTTPException(status_code=400, detail="Zuschnitt zu klein")

    edited = await asyncio.to_thread(
        imaging.edit_jpeg, original.read_bytes(), rotate, angle, crop
    )
    current.write_bytes(edited)
    return card_dict(_card_or_404(card_id))


@app.post("/api/cards/{card_id}/back", dependencies=[auth])
async def add_back(card_id: str, file: UploadFile) -> dict:
    """Rueckseite nachtragen. Sie wird ausgelesen, fuellt aber nur Felder, die
    die Vorderseite offen gelassen hat - dort steht oft nur die Mobilnummer."""
    row = _card_or_404(card_id)
    raw = await file.read()
    if not raw:
        raise HTTPException(status_code=400, detail="Leere Datei")
    if len(raw) > MAX_UPLOAD_BYTES:
        raise HTTPException(status_code=413, detail="Foto ist groesser als 30 MB")

    try:
        _, encoded, _ = await asyncio.to_thread(_process_photo, raw)
    except Exception as exc:
        log.exception("Rueckseite konnte nicht verarbeitet werden")
        raise HTTPException(status_code=400, detail=f"Bild nicht lesbar: {exc}") from exc
    if not encoded:
        raise HTTPException(status_code=400, detail="Keine Karte im Bild gefunden")

    name = f"{card_id}-back.jpg"
    (config.IMAGE_DIR / name).write_bytes(encoded[0])
    db.execute(
        "UPDATE cards SET back_image_file = ?, extract_status = 'offen' WHERE id = ?",
        (name, card_id),
    )
    start_reading([card_id], side="back")
    return card_dict(_card_or_404(card_id))


# ---------------------------------------------------------------------------- Export

@app.get("/vcf/{card_id}", dependencies=[auth])
def download_vcard(card_id: str) -> Response:
    card = dict(_card_or_404(card_id))
    db.execute("UPDATE cards SET exported_at = ? WHERE id = ?", (now_iso(), card_id))
    return Response(
        content=vcard.build(card),
        media_type="text/vcard; charset=utf-8",
        headers={"Content-Disposition": f'attachment; filename="{vcard.filename(card)}"'},
    )


# ----------------------------------------------------------------------- Erinnerungen

@app.get("/api/reminders", dependencies=[auth])
def list_reminders() -> dict:
    rows = db.query(
        """
        SELECT r.*, c.first_name, c.last_name, c.company
        FROM reminders r JOIN cards c ON c.id = r.card_id
        ORDER BY r.sent_at IS NOT NULL, r.due_at
        """
    )
    return {"reminders": [dict(r) for r in rows], "mail_configured": bool(config.SMTP_HOST)}


@app.post("/api/cards/{card_id}/reminders", dependencies=[auth])
async def create_reminder(card_id: str, request: Request) -> dict:
    _card_or_404(card_id)
    payload = await request.json()
    text = (payload.get("text") or "").strip()
    due_at = (payload.get("due_at") or "").strip()
    if not text:
        raise HTTPException(status_code=400, detail="Erinnerungstext fehlt")
    try:
        parsed = datetime.fromisoformat(due_at.replace("Z", "+00:00"))
    except ValueError as exc:
        raise HTTPException(status_code=400, detail="Datum nicht lesbar") from exc
    if parsed.tzinfo is None:
        parsed = parsed.replace(tzinfo=timezone.utc)

    reminder_id = uuid.uuid4().hex
    db.execute(
        "INSERT INTO reminders (id, card_id, text, due_at, created_at) VALUES (?, ?, ?, ?, ?)",
        (reminder_id, card_id, text, parsed.astimezone(timezone.utc).isoformat(), now_iso()),
    )
    return dict(db.query_one("SELECT * FROM reminders WHERE id = ?", (reminder_id,)))


@app.delete("/api/reminders/{reminder_id}", dependencies=[auth])
def delete_reminder(reminder_id: str) -> dict:
    db.execute("DELETE FROM reminders WHERE id = ?", (reminder_id,))
    return {"ok": True}


# --------------------------------------------------------------------------- Statisch

@app.get("/")
def index() -> FileResponse:
    return FileResponse(STATIC_DIR / "index.html")


@app.get("/sw.js")
def service_worker() -> FileResponse:
    # Muss von der Wurzel kommen, sonst gilt der Service Worker nur fuer /static.
    return FileResponse(STATIC_DIR / "sw.js", media_type="application/javascript")


@app.get("/manifest.webmanifest")
def manifest() -> FileResponse:
    return FileResponse(STATIC_DIR / "manifest.webmanifest", media_type="application/manifest+json")


@app.get("/robots.txt")
def robots() -> PlainTextResponse:
    return PlainTextResponse("User-agent: *\nDisallow: /\n")


app.mount("/static", StaticFiles(directory=STATIC_DIR), name="static")