Kansho/backend/provider_settings.py

"""Admin-configurable generate/detect endpoints. Keys stay in env, never in the DB."""
from __future__ import annotations

import json
import os
import uuid
from pathlib import Path
from urllib.parse import urlparse

from db import get_db
from detect_learning import get_detect_operating_mode
from env_loader import (
    allows_remote_plaintext_detect,
    remote_plaintext_detect_reason,
    runtime_env,
    upsert_env_value,
)

SEED_PATH = Path(__file__).resolve().parent / "config" / "providers.seed.json"
ROLES = ("generate", "detect")
OLLAMA_LAN_ID = "llm-ollama-lan"
OLLAMA_LAN_URL = "http://192.168.2.144:11434/v1/chat/completions"
_KEY_ENV = {
    "generate": "KANSHO_PROVIDER_KEY",
    "detect": "KANSHO_DETECT_PROVIDER_KEY",
}
# Interne Rollen bleiben generate/detect. Fachlich: Sprachmodell vs. Maskierung.
ROLE_META = {
    "generate": {
        "title": "Sprachmodell",
        "task": "Dialogzug und Journalentwurf. Nur maskierter Kontext. Wählt die Operation im Dialogzug mit.",
    },
    "detect": {
        "title": "Maskierung",
        "task": "Eigener Detect-Endpunkt und eigenes Modell. Vollständige semantische Detection des persönlichen Egress. Ziel: lokales Detect-Modell (Ollama). Übergang: externes Klartext-Detect nur mit Operator-Freigabe. Kein Pattern-Fallback.",
    },
}


class SettingsError(Exception):
    def __init__(self, code: str, message: str, status_code: int = 400):
        super().__init__(message)
        self.code = code
        self.message = message
        self.status_code = status_code


def _validate_url(url: str) -> str:
    url = (url or "").strip()
    if not url:
        return ""
    parsed = urlparse(url)
    if parsed.scheme not in {"http", "https"} or not parsed.netloc:
        raise SettingsError("provider_url_invalid", "URL muss http(s) sein.")
    return url


def _profile_row(conn, profile_id: str) -> dict | None:
    row = conn.execute("SELECT * FROM llm_profiles WHERE id = ?", (profile_id,)).fetchone()
    return dict(row) if row else None


def _public_profile(row: dict) -> dict:
    from providers import is_local_url

    url = row.get("url") or ""
    return {
        "id": row.get("id") or "",
        "title": row.get("title") or "",
        "name": row.get("name") or "",
        "url": url,
        "model": row.get("model") or "",
        "zdr": bool(row.get("zdr", 1)),
        "no_train": bool(row.get("no_train", 1)),
        "local": is_local_url(url),
        "updated": row.get("updated") or "",
    }


def seed_provider_settings(conn) -> None:
    items = json.loads(SEED_PATH.read_text(encoding="utf-8"))
    for item in items:
        role = item["role"]
        if role not in ROLES:
            continue
        existing = conn.execute("SELECT role FROM provider_settings WHERE role = ?", (role,)).fetchone()
        if existing:
            continue
        conn.execute(
            """
            INSERT INTO provider_settings (role, name, url, model, zdr, no_train)
            VALUES (?, ?, ?, ?, ?, ?)
            """,
            (
                role,
                item.get("name") or "",
                item.get("url") or "",
                item.get("model") or "",
                1 if item.get("zdr", True) else 0,
                1 if item.get("no_train", True) else 0,
            ),
        )
    seed_llm_profiles(conn)


def seed_llm_profiles(conn) -> None:
    """Named presets. Existing rows are not overwritten. Keys never stored."""
    ollama = conn.execute("SELECT * FROM llm_profiles WHERE id = ?", (OLLAMA_LAN_ID,)).fetchone()
    if not ollama:
        conn.execute(
            """
            INSERT INTO llm_profiles (id, title, name, url, model, zdr, no_train)
            VALUES (?, ?, ?, ?, ?, 1, 1)
            """,
            (OLLAMA_LAN_ID, "Ollama (LAN)", "ollama", OLLAMA_LAN_URL, "phi3:mini"),
        )
    elif not (dict(ollama).get("model") or "").strip():
        conn.execute(
            "UPDATE llm_profiles SET model = ?, updated = datetime('now') WHERE id = ?",
            ("phi3:mini", OLLAMA_LAN_ID),
        )
    for role in ROLES:
        row = conn.execute("SELECT * FROM provider_settings WHERE role = ?", (role,)).fetchone()
        if not row:
            continue
        data = dict(row)
        profile_id = (data.get("profile_id") or "").strip()
        if profile_id and _profile_row(conn, profile_id):
            continue
        url = (data.get("url") or "").strip()
        model = (data.get("model") or "").strip()
        if not url:
            continue
        match = conn.execute(
            "SELECT id FROM llm_profiles WHERE url = ? AND model = ? ORDER BY created LIMIT 1",
            (url, model),
        ).fetchone()
        if match:
            chosen = match["id"]
        else:
            chosen = f"llm-{role}-current"
            if _profile_row(conn, chosen):
                chosen = str(uuid.uuid4())
            title = "Sprachmodell (aktuell)" if role == "generate" else "Maskierung (aktuell)"
            conn.execute(
                """
                INSERT INTO llm_profiles (id, title, name, url, model, zdr, no_train)
                VALUES (?, ?, ?, ?, ?, ?, ?)
                """,
                (
                    chosen,
                    title,
                    data.get("name") or "",
                    url,
                    model,
                    1 if data.get("zdr", 1) else 0,
                    1 if data.get("no_train", 1) else 0,
                ),
            )
        conn.execute(
            "UPDATE provider_settings SET profile_id = ? WHERE role = ?",
            (chosen, role),
        )


def list_profiles() -> list[dict]:
    with get_db() as conn:
        rows = conn.execute("SELECT * FROM llm_profiles ORDER BY title, created").fetchall()
    return [_public_profile(dict(row)) for row in rows]


def get_profile(profile_id: str) -> dict | None:
    with get_db() as conn:
        row = _profile_row(conn, profile_id)
    return _public_profile(row) if row else None


def _write_profile(
    conn,
    profile_id: str,
    *,
    title: str,
    name: str,
    url: str,
    model: str,
    zdr: bool,
    no_train: bool,
) -> dict:
    title = (title or "").strip()
    if not title:
        raise SettingsError("profile_title_required", "Ein Profil braucht einen Namen.")
    url = _validate_url(url)
    if not url:
        raise SettingsError("provider_url_required", "Ein Profil braucht eine URL.")
    model = (model or "").strip()
    name = (name or "").strip()
    existing = _profile_row(conn, profile_id)
    if existing:
        conn.execute(
            """
            UPDATE llm_profiles
            SET title = ?, name = ?, url = ?, model = ?, zdr = ?, no_train = ?, updated = datetime('now')
            WHERE id = ?
            """,
            (title, name, url, model, 1 if zdr else 0, 1 if no_train else 0, profile_id),
        )
    else:
        conn.execute(
            """
            INSERT INTO llm_profiles (id, title, name, url, model, zdr, no_train)
            VALUES (?, ?, ?, ?, ?, ?, ?)
            """,
            (profile_id, title, name, url, model, 1 if zdr else 0, 1 if no_train else 0),
        )
    for role in ROLES:
        active = conn.execute(
            "SELECT role FROM provider_settings WHERE role = ? AND profile_id = ?",
            (role, profile_id),
        ).fetchone()
        if not active:
            continue
        conn.execute(
            """
            UPDATE provider_settings
            SET name = ?, url = ?, model = ?, zdr = ?, no_train = ?, updated = datetime('now')
            WHERE role = ?
            """,
            (name, url, model, 1 if zdr else 0, 1 if no_train else 0, role),
        )
    row = _profile_row(conn, profile_id)
    return _public_profile(row or {})


def create_profile(*, title: str, name: str, url: str, model: str, zdr: bool, no_train: bool) -> dict:
    profile_id = str(uuid.uuid4())
    with get_db() as conn:
        return _write_profile(
            conn,
            profile_id,
            title=title,
            name=name,
            url=url,
            model=model,
            zdr=zdr,
            no_train=no_train,
        )


def update_profile(
    profile_id: str,
    *,
    title: str,
    name: str,
    url: str,
    model: str,
    zdr: bool,
    no_train: bool,
) -> dict:
    with get_db() as conn:
        if not _profile_row(conn, profile_id):
            raise SettingsError("profile_missing", "Profil fehlt.", 404)
        return _write_profile(
            conn,
            profile_id,
            title=title,
            name=name,
            url=url,
            model=model,
            zdr=zdr,
            no_train=no_train,
        )


def delete_profile(profile_id: str) -> None:
    with get_db() as conn:
        if not _profile_row(conn, profile_id):
            raise SettingsError("profile_missing", "Profil fehlt.", 404)
        used = conn.execute(
            "SELECT role FROM provider_settings WHERE profile_id = ?",
            (profile_id,),
        ).fetchall()
        if used:
            roles = ", ".join(row["role"] for row in used)
            raise SettingsError(
                "profile_in_use",
                f"Profil ist noch der Stufe {roles} zugeordnet. Zuerst ein anderes Profil wählen.",
            )
        conn.execute("DELETE FROM llm_profiles WHERE id = ?", (profile_id,))


def activate_profile(role: str, profile_id: str) -> dict:
    if role not in ROLES:
        raise SettingsError("unknown_provider_role", "Unbekannte Provider-Rolle.")
    with get_db() as conn:
        row = _profile_row(conn, profile_id)
        if not row:
            raise SettingsError("profile_missing", "Profil fehlt.", 404)
        url = (row.get("url") or "").strip()
        model = (row.get("model") or "").strip()
        if role == "generate" and not url:
            raise SettingsError("provider_url_required", "Das Sprachmodell braucht eine URL.")
        if role == "detect" and not url:
            raise SettingsError("provider_url_required", "Die Maskierung braucht eine URL.")
        if not model:
            raise SettingsError("provider_model_required", "Das Profil braucht ein Modell.")
        conn.execute(
            """
            INSERT INTO provider_settings (role, name, url, model, zdr, no_train, profile_id, updated)
            VALUES (?, ?, ?, ?, ?, ?, ?, datetime('now'))
            ON CONFLICT(role) DO UPDATE SET
              name = excluded.name,
              url = excluded.url,
              model = excluded.model,
              zdr = excluded.zdr,
              no_train = excluded.no_train,
              profile_id = excluded.profile_id,
              updated = datetime('now')
            """,
            (
                role,
                row.get("name") or "",
                url,
                model,
                1 if row.get("zdr", 1) else 0,
                1 if row.get("no_train", 1) else 0,
                profile_id,
            ),
        )
    return get_setting(role) or {}


def get_setting(role: str) -> dict | None:
    if role not in ROLES:
        return None
    with get_db() as conn:
        row = conn.execute("SELECT * FROM provider_settings WHERE role = ?", (role,)).fetchone()
    return dict(row) if row else None


def upsert_setting(
    role: str,
    *,
    name: str,
    url: str,
    model: str,
    zdr: bool,
    no_train: bool,
    profile_id: str | None = None,
    save_as_title: str | None = None,
) -> dict:
    if role not in ROLES:
        raise SettingsError("unknown_provider_role", "Unbekannte Provider-Rolle.")
    url = _validate_url(url)
    model = (model or "").strip()
    name = (name or "").strip()
    if role == "generate" and not url:
        raise SettingsError("provider_url_required", "Das Sprachmodell braucht eine URL.")
    if role == "generate" and not model:
        raise SettingsError("provider_model_required", "Das Sprachmodell braucht ein Modell.")
    title = (save_as_title or "").strip()
    with get_db() as conn:
        current = conn.execute("SELECT * FROM provider_settings WHERE role = ?", (role,)).fetchone()
        current_id = (dict(current).get("profile_id") if current else "") or ""
        chosen = (profile_id or "").strip() or current_id
        if title:
            chosen = str(uuid.uuid4())
            _write_profile(
                conn,
                chosen,
                title=title,
                name=name,
                url=url,
                model=model,
                zdr=zdr,
                no_train=no_train,
            )
        elif chosen:
            existing = _profile_row(conn, chosen)
            if not existing:
                raise SettingsError("profile_missing", "Profil fehlt.", 404)
            _write_profile(
                conn,
                chosen,
                title=existing.get("title") or title or role,
                name=name,
                url=url,
                model=model,
                zdr=zdr,
                no_train=no_train,
            )
        conn.execute(
            """
            INSERT INTO provider_settings (role, name, url, model, zdr, no_train, profile_id, updated)
            VALUES (?, ?, ?, ?, ?, ?, ?, datetime('now'))
            ON CONFLICT(role) DO UPDATE SET
              name = excluded.name,
              url = excluded.url,
              model = excluded.model,
              zdr = excluded.zdr,
              no_train = excluded.no_train,
              profile_id = excluded.profile_id,
              updated = datetime('now')
            """,
            (role, name, url, model, 1 if zdr else 0, 1 if no_train else 0, chosen or None),
        )
    return get_setting(role) or {}


def set_role_key(role: str, key: str) -> None:
    if role not in ROLES:
        raise SettingsError("unknown_provider_role", "Unbekannte Provider-Rolle.")
    value = (key or "").strip()
    if not value:
        raise SettingsError("provider_key_empty", "Leerer Key wird nicht gespeichert.")
    upsert_env_value(_KEY_ENV[role], value)


def public_role_status(role: str) -> dict:
    from providers import detect_provider, generate_provider, is_local_url

    row = get_setting(role) or {}
    config = generate_provider() if role == "generate" else detect_provider()
    local = is_local_url(row.get("url") or "")
    own_key = bool((os.environ.get(_KEY_ENV[role]) or "").strip())
    generate_key = bool((os.environ.get(_KEY_ENV["generate"]) or "").strip())
    if own_key:
        key_source = "own"
    elif role == "detect" and generate_key and not local:
        key_source = "generate"
    else:
        key_source = "none"
    meta = ROLE_META.get(role, {})
    return {
        "role": role,
        "title": meta.get("title") or role,
        "task": meta.get("task") or "",
        "name": row.get("name") or (config.name if config else ""),
        "url": row.get("url") or "",
        "model": row.get("model") or "",
        "zdr": bool(row.get("zdr", 1)),
        "no_train": bool(row.get("no_train", 1)),
        "profile_id": row.get("profile_id") or "",
        "local": local,
        "key_present": key_source != "none",
        "key_source": key_source,
        "ready": config is not None,
        "mode": config.mode if config else "unconfigured",
        "remote_plaintext_detect": bool(config and not config.local and config.mode == "http" and role == "detect"),
        "remote_plaintext_allowed": allows_remote_plaintext_detect() if role == "detect" else None,
        "remote_plaintext_reason": remote_plaintext_detect_reason() if role == "detect" else None,
        "updated": row.get("updated") or "",
    }


def public_status() -> dict:
    return {
        "roles": [public_role_status(role) for role in ROLES],
        "profiles": list_profiles(),
        "runtime_env": runtime_env(),
        "note": "LLM-Profile speichern URL und Modell je Endpunkt. Maskierung und Sprachmodell wählen je ein Profil. Semantische Detection ist Pflicht vor Generate. LAN-Ollama gilt als lokal. Keys nur in .env.",
        "remote_plaintext_reason": remote_plaintext_detect_reason(),
        "remote_plaintext_allowed": allows_remote_plaintext_detect(),
        "ollama_url": OLLAMA_LAN_URL,
        "detect_operating_mode": get_detect_operating_mode(),
    }