bifroest/app/demo/__init__.py

"""Demo-Workspaces für Vorführungen und den PoC (`aiw demo`): Repository, Datei zur Analyse, Dokumentenordner,
Beispielfragen. Jede Demo legt alles reproduzierbar an (gleiche Daten bei jedem Lauf); die Inhalte stehen in
den Modulen dieses Pakets, hier nur der gemeinsame Ablauf.
"""

import io
import os
import shutil
import subprocess
import tempfile
from collections.abc import Callable
from dataclasses import dataclass
from datetime import date, datetime, timedelta
from pathlib import Path

from sqlalchemy import select

from app.core.config import Settings
from app.core.crypto import Vault
from app.core.database import Database

AUTHORS = {"ben": ("Ben Entwickler", "ben@demo.example"), "clara": ("Clara Fachbereich", "clara@demo.example"),
           "anna": ("Anna Admin", "anna@demo.example")}

# Commit: (Tage zurück, Autor, Nachricht, {Pfad: Inhalt}); Branch: (Name, Tage zurück, Autor, Nachricht, Dateien),
# abgezweigt vom vorletzten Commit auf main
Commit = tuple[int, str, str, dict[str, str]]


def day(days: int) -> date:
    """Datum relativ zu heute: Protokolle passen so zeitlich zu den Commits, egal wann die Demo angelegt wird."""
    return date.today() - timedelta(days=days)


@dataclass(frozen=True)
class Demo:
    key: str
    name: str
    description: str
    examples: list[str]
    repo: str
    commits: list[Commit]
    branch: tuple[str, int, str, str, dict[str, str]]
    upload: tuple[str, Callable[[], bytes]]
    folder: str
    documents: dict[str, str]
    # Zusätzlich anbinden (z. B. Beispieldatenbank): (db, settings, user_id, workspace_id), auch bei vorhandenem
    # Workspace
    extra: Callable[[Database, Settings, int, int], None] | None = None


def git(cwd: Path, *args: str, env: dict | None = None) -> None:
    subprocess.run(["git", *args], cwd=cwd, check=True, capture_output=True, env={**os.environ, **(env or {})})


def write_files(root: Path, files: dict[str, str]) -> None:
    for path, content in files.items():
        target = root / path
        target.parent.mkdir(parents=True, exist_ok=True)
        target.write_text(content)


def commit(work: Path, days: int, author: str, message: str) -> None:
    name, email = AUTHORS[author]
    when = (datetime.now().astimezone() - timedelta(days=days)).replace(hour=10, minute=15, second=0,
                                                                        microsecond=0).isoformat()
    git(work, "add", "-A")
    git(work, "commit", "-q", "-m", message, env={
        "GIT_AUTHOR_NAME": name, "GIT_AUTHOR_EMAIL": email, "GIT_AUTHOR_DATE": when,
        "GIT_COMMITTER_NAME": name, "GIT_COMMITTER_EMAIL": email, "GIT_COMMITTER_DATE": when})


def build_repo(demo: Demo, target: Path) -> None:
    """Bare-Repository mit Verlauf und einem Feature-Branch."""
    if target.exists():
        shutil.rmtree(target)
    target.parent.mkdir(parents=True, exist_ok=True)
    with tempfile.TemporaryDirectory() as tmp:
        work = Path(tmp) / "work"
        work.mkdir()
        git(work, "init", "-q", "-b", "main")
        for days, author, message, files in demo.commits:
            write_files(work, files)
            commit(work, days, author, message)
        branch, days, author, message, files = demo.branch
        git(work, "checkout", "-q", "-b", branch, "main~1")
        write_files(work, files)
        commit(work, days, author, message)
        git(work, "checkout", "-q", "main")
        git(work.parent, "clone", "-q", "--bare", str(work), str(target.resolve()))


def repo_path(settings: Settings, demo: Demo) -> Path:
    return Path(settings.git_cache_dir).parent / "demo" / f"{demo.repo}.git"


def all_demos() -> dict[str, Demo]:
    from app.demo import bestellservice, mcp
    return {d.key: d for d in (bestellservice.DEMO, mcp.DEMO)}


def run(settings: Settings, keys: list[str] | None = None, reset: bool = False) -> int:
    from app.cli import seed_demo
    from app.identity.models import User
    from app.identity.service import principal_for
    from app.tools import service as tools

    demos = all_demos()
    chosen = [demos[k] for k in keys] if keys else list(demos.values())
    seed_demo(settings)
    db = Database(settings.database_url)
    try:
        with db.session() as session:
            anna = session.scalar(select(User).where(User.username == "anna"))
            classified = tools.classify_suggested(session, principal_for(session, anna.id))
        if classified:
            print(f"{classified} eingebaute Tools nach Vorschlag eingestuft (wie „Vorschläge übernehmen“ unter Tools).")
        for demo in chosen:
            create(db, settings, demo, anna.id, reset)
        print("Fertig. Anmelden als anna, ben oder clara (Passwort AIW_DEMO_PASSWORD).")
        return 0
    finally:
        db.dispose()


def create(db: Database, settings: Settings, demo: Demo, user_id: int, reset: bool) -> None:
    from app.ai import service as ai
    from app.data_sources import service as sources
    from app.documents import service as documents
    from app.identity.models import Team
    from app.identity.service import principal_for
    from app.uploads import service as uploads
    from app.workspaces import removal
    from app.workspaces import service as workspaces
    from app.workspaces.models import Workspace

    vault = Vault(settings.secret_key)
    with db.session() as session:
        team = session.scalar(select(Team).where(Team.slug == "demo"))
        existing = session.scalar(select(Workspace).where(Workspace.team_id == team.id, Workspace.name == demo.name,
                                                          Workspace.deleted.is_(None)))
        if existing is not None and not reset:
            print(f"„{demo.name}“ existiert schon (ID {existing.id}). Neu anlegen mit: aiw demo {demo.key} --reset")
            existing_id = existing.id
        else:
            existing_id = None
        if existing is not None and reset:
            removal.remove(session, settings, existing, None)
            print(f"Alten Workspace „{demo.name}“ gelöscht.")
    if existing_id is not None:
        if demo.extra:
            demo.extra(db, settings, user_id, existing_id)
        return
    with db.session() as session:
        principal = principal_for(session, user_id)
        workspace = workspaces.create(session, principal, team.id, demo.name, demo.description)
        workspace_id = workspace.id
        for username in ("ben", "clara"):
            workspaces.set_member(session, principal, workspace_id, username, "member")
        workspaces.set_examples(session, principal, workspace_id, "\n".join(demo.examples))
        model = ai.effective_model(session, workspace)
    print(f"„{demo.name}“ angelegt (ID {workspace_id}), Mitglieder anna (Admin), ben, clara.")

    # Repository
    if settings.git_allow_local:
        target = repo_path(settings, demo)
        build_repo(demo, target)
        with db.session() as session:
            source = sources.create_git(session, principal, settings, vault, workspace_id, name=demo.repo,
                                        url=target.resolve().as_uri(), branch="main", token="",
                                        exposure_confirmed=True)
            source_id = source.id
        with db.session() as session:
            source = sources.require_testable(session, principal, source_id)
        result = sources.run_test(source, settings, vault)
        with db.session() as session:
            sources.save_test(session, principal, source_id, result)
        print(f"  Repository {demo.repo} ({len(demo.commits)} Commits auf main, Branch {demo.branch[0]}): "
              f"{'verbunden' if result.ok else 'Fehler: ' + result.message}")
    else:
        print("  Repository übersprungen: lokale Git-Quellen sind aus (AIW_GIT_ALLOW_LOCAL=true für die Demo).")

    # Datei zur Analyse
    filename, generate = demo.upload
    with db.session() as session:
        upload = uploads.save(session, principal, settings, workspace_id, filename, io.BytesIO(generate()))
        print(f"  Datei {filename}: {upload.row_count} Zeilen als Tabelle {upload.table_name}.")

    # Dokumentenordner (Textdateien, brauchen keine Dokumenten-Auslese)
    with db.session() as session:
        folder = documents.create(session, principal, workspace_id, demo.folder, documents.WORKSPACE, True)
        folder_id = folder.id
    modified = int(datetime.now().timestamp() * 1000)
    with db.session() as session:
        for path, text in demo.documents.items():
            documents.put(session, principal, settings, folder_id, path, modified, io.BytesIO(text.encode()))
        documents.prune(session, principal, settings, folder_id, list(demo.documents))
    print(f"  Ordner {demo.folder}: {len(demo.documents)} Dokumente (für den ganzen Workspace).")
    if demo.extra:
        demo.extra(db, settings, user_id, workspace_id)
    if model is None:
        print("  Hinweis: Noch kein Modell eingerichtet; das macht ein Plattform-Admin unter „Modelle und Zugänge“.")