shopdb-flask/plugins/usb/cli.py

"""CLI: move USB records off an external cmmc_usb database and into ShopDB.

    flask usb import-external              report what would happen, change nothing
    flask usb import-external --commit     write it

A site running the external mode keeps its devices and its check-in/out history
in a separate cmmc_usb database. This reads that database through the same
connection settings the routes already use, converts it to the self-hosted
tables, and leaves the source untouched. After a clean run, set
usb_directory_mode to selfhosted and the external database can be retired.

TWO THINGS THIS HAS TO RECONCILE, and they are the reason it is a command
rather than a few INSERTs:

The two schemas disagree about shape. External is an EVENT LOG - one row per
action, check-out or check-in - while self-hosted keeps a PAIRED ROW, one
checkout carrying both its times. Pairing is per device and in time order, and
real logs are not tidy: a check-out with no return is an open checkout, but a
return with no matching check-out, or two check-outs in a row, mean something
was recorded wrong years ago. Those are reported rather than guessed at, and
they never silently vanish.

Identity is the other one. usbdevices.serialnumber is unique and NOT NULL, and
the external devices table has no serial column - only device_id. So device_id
becomes the serial. A blank or duplicated device_id stops the import for that
device instead of inventing a key.

Dry run by default. The point of the first run is the report.
"""

from collections import defaultdict

import click
from flask.cli import AppGroup, with_appcontext

from shopdb.api import db, cmmc_usb_connection

from .models import USBDevice, USBCheckout


usb_cli = AppGroup('usb', help='USB device records.')


# The external status vocabulary, mapped to the self-hosted boolean. Anything
# unrecognised is treated as checked in and reported, rather than assumed.
_CHECKED_OUT = {'checked-out', 'checked out', 'out', 'issued'}
_RETIRED = {'retired', 'disposed', 'destroyed'}


def _external_rows():
    """(devices, log) from the external database, oldest log entry first."""
    conn = cmmc_usb_connection()
    try:
        with conn.cursor() as cur:
            cur.execute('SELECT device_id, device_desc, device_owner, status, '
                        'locker_location FROM devices')
            devices = cur.fetchall()
            cur.execute('SELECT log_id, badge_number, device_id, action, timestamp, '
                        'scanned_viruses, locker_location, sanitized '
                        'FROM checkinoutlog ORDER BY device_id, timestamp, log_id')
            log = cur.fetchall()
    finally:
        conn.close()
    return devices, log


def _is_checkout(action):
    return str(action or '').strip().lower().replace('_', '-') in (
        'check-out', 'checkout', 'out')


def _is_checkin(action):
    return str(action or '').strip().lower().replace('_', '-') in (
        'check-in', 'checkin', 'in', 'return')


def _pair(entries, problems, deviceid):
    """Walk one device's log in time order and fold it into checkout pairs."""
    pairs, open_row = [], None
    for row in entries:
        action = row.get('action')
        if _is_checkout(action):
            if open_row is not None:
                # Two check-outs with no return between them. The first is
                # unclosed and stays that way; say so rather than inventing a
                # return time that nobody recorded.
                problems.append(
                    'device %s: check-out at %s while %s was still out - first '
                    'left open' % (deviceid, row.get('timestamp'),
                                   open_row.get('badge_number')))
                pairs.append((open_row, None))
            open_row = row
        elif _is_checkin(action):
            if open_row is None:
                problems.append(
                    'device %s: check-in at %s with no matching check-out - '
                    'skipped' % (deviceid, row.get('timestamp')))
                continue
            pairs.append((open_row, row))
            open_row = None
        else:
            problems.append('device %s: unrecognised action %r at %s - skipped'
                            % (deviceid, action, row.get('timestamp')))
    if open_row is not None:
        pairs.append((open_row, None))
    return pairs


@usb_cli.command('import-external')
@click.option('--commit', is_flag=True, default=False,
              help='Write the records. Without this the command only reports.')
@with_appcontext
def import_external(commit):
    """Copy devices and check-in/out history from the external cmmc_usb DB."""
    try:
        devices, log = _external_rows()
    except Exception as error:
        raise click.ClickException(
            'Could not read the external database: %s\n'
            'Check the cmmc_usb_db_* settings, and that this host can reach it.'
            % error)

    problems = []
    seen_serials = set()
    made_devices = 0
    made_checkouts = 0
    open_checkouts = 0

    by_device = defaultdict(list)
    for row in log:
        by_device[row.get('device_id')].append(row)

    for source in devices:
        serial = (source.get('device_id') or '').strip()
        if not serial:
            problems.append('device with no device_id - skipped, it cannot be '
                            'given a serial number')
            continue
        if serial in seen_serials:
            problems.append('device_id %r appears twice - only the first is '
                            'imported' % serial)
            continue
        seen_serials.add(serial)

        status = str(source.get('status') or '').strip().lower()
        if status and status not in _CHECKED_OUT and status not in _RETIRED \
                and status not in ('checked-in', 'checked in', 'in', 'available'):
            problems.append('device %s: unrecognised status %r - treated as '
                            'checked in' % (serial, source.get('status')))

        device = USBDevice.query.filter_by(serialnumber=serial).first()
        if device is None:
            device = USBDevice(serialnumber=serial)
            db.session.add(device)
            made_devices += 1
        device.label = source.get('device_desc') or device.label
        device.storagelocation = source.get('locker_location') or device.storagelocation
        device.notes = _append_owner(device.notes, source.get('device_owner'))
        device.ischeckedout = status in _CHECKED_OUT

        db.session.flush()

        for out_row, in_row in _pair(by_device.get(source.get('device_id'), []),
                                     problems, serial):
            checkout = USBCheckout(
                usbdeviceid=device.usbdeviceid,
                sso=str(out_row.get('badge_number') or '')[:20],
                checkouttime=out_row.get('timestamp'),
                checkintime=in_row.get('timestamp') if in_row else None,
            )
            if in_row:
                checkout.waswiped = _tri(in_row.get('sanitized'))
                checkout.scannedviruses = _tri(in_row.get('scanned_viruses'))
            else:
                open_checkouts += 1
            db.session.add(checkout)
            made_checkouts += 1

        # The device's own flag and the history have to agree, or the kiosk
        # will offer a drive that the log says somebody still holds.
        if device.ischeckedout and not _has_open(device):
            problems.append('device %s: marked checked out, but its log has no '
                            'open check-out' % serial)

    click.echo('devices read from external : %d' % len(devices))
    click.echo('log rows read              : %d' % len(log))
    click.echo('devices to create          : %d' % made_devices)
    click.echo('checkouts to create        : %d' % made_checkouts)
    click.echo('  of which still open      : %d' % open_checkouts)

    if problems:
        click.echo('')
        click.echo('%d row(s) need a human decision:' % len(problems))
        for line in problems:
            click.echo('  - %s' % line)

    if not commit:
        db.session.rollback()
        click.echo('')
        click.echo('DRY RUN - nothing was written. Re-run with --commit when the '
                   'report above looks right.')
        return

    db.session.commit()
    click.echo('')
    click.echo('Written. Next: set usb_directory_mode to selfhosted in Settings, '
               'confirm the kiosk and the device list against the external '
               'database, and only then retire it.')


def _tri(value):
    """External flags are 0/1/NULL. Keep NULL as None: unrecorded is not false."""
    if value is None:
        return None
    return bool(value)


def _has_open(device):
    return USBCheckout.query.filter_by(usbdeviceid=device.usbdeviceid,
                                       checkintime=None).count() > 0


def _append_owner(notes, owner):
    """External device_owner has no column here, so keep it rather than lose it."""
    owner = (owner or '').strip()
    if not owner:
        return notes
    line = 'Imported owner: %s' % owner
    if notes and line in notes:
        return notes
    return (notes + '\n' + line) if notes else line


def build_cli():
    return [usb_cli]