def main(opts=None, cli_args=None, log_stream=sys.stdout):
if opts is None:
opts = argparse_get_parser()\
.parse_args(sys.argv[1:] if cli_args is None else cli_args)
else:
parser = None # to check and re-raise django CommandError
if not isinstance(opts, argparse.Namespace): opts = argparse.Namespace(**opts)
assert cli_args is None, cli_args
command_logger_setup( log, opts,
stream=log_stream, verbose_level=logging.EXTRA )
# Process --interval-parameters, --commit-interval
try:
if isinstance(opts.interval_parameters, types.StringTypes):
params = cli_defaults['interval_parameters'].copy()
for v in opts.interval_parameters.split(':'):
k, vs = v.split('=')
if k not in params:
raise CommandError('Unrecognized interval parameter: {0}'.format(k))
if vs.lower() == 'true': v = True
elif vs.lower() == 'false': v = False
elif vs.lower() == 'none': v = 0
else:
try: v = float(vs.rstrip('sdh'))
except ValueError:
raise CommandError('Unrecognized interval parameter value: {0}'.format(vs))
if vs.endswith('h'): v /= float(24)
elif vs.endswith('s'): v /= float(3600 * 24)
params[k] = v
opts.interval_parameters = params
if opts.commit_interval:
if opts.commit_interval.isdigit():
opts.commit_interval = int(opts.commit_interval)
elif opts.commit_interval.endswith('s'):
opts.commit_interval = timedelta(0, int(opts.commit_interval[:-1]))
else:
raise CommandError( 'Invalid'
' interval value: {0}'.format(opts.commit_interval) )
if opts.report_after:
try: v = float(opts.report_after.rstrip('sdh'))
except ValueError:
raise CommandError('Unrecognized timespan value: {0}'.format(opts.report_after))
if opts.report_after.endswith('h'): v /= float(24)
elif opts.report_after.endswith('s'): v /= float(3600 * 24)
opts.report_after = timedelta(v)
except CommandError as err:
if not parser: raise
parser.error(*err.args)
# Make sure logging won't choke on encoding
import codecs
codec = codecs.getwriter('utf-8')
sys.stdout = codec(sys.stdout)
sys.stderr = codec(sys.stderr)
bulk_update(opts)
def handle(self, **opts):
opts = type(
b'Opts', (object, ),
dict((k.replace('-', '_'), v) for k, v in opts.viewitems()))
log = logging.getLogger('feedjack.purge')
command_logger_setup(log, opts, stream=self.stdout)
if not opts.feed and not opts.site:
feeds = set(models.Feed.all())
log.info('All feeds will be affected (%s)', len(feeds))
else:
feeds = set()
if opts.feed:
try:
feeds.update(
models.Feed.objects.get_by_string(name)
for name in opts.feed)
except (models.ObjectDoesNotExist,
models.MultipleObjectsReturned) as err:
raise CommandError(unicode(err))
if opts.site:
try:
sites = list(
models.Site.objects.get_by_string(name)
for name in opts.site)
except (models.ObjectDoesNotExist,
models.MultipleObjectsReturned) as err:
raise CommandError(unicode(err))
for site in sites:
feeds.update(site.feeds)
if log.isEnabledFor(logging.INFO):
if log.isEnabledFor(logging.DEBUG):
log.debug('List of affected feeds (%s):', len(feeds))
for feed in sorted(feeds, key=op.attrgetter('pk')):
log.debug(' - [%s] %s', feed.pk, feed)
else:
log.info(
'Number of affected feeds: %s (listed with more verbose logging)',
len(feeds))
if opts.type == 'by-age':
ts0, ts = timezone.localtime(timezone.now()), parse_timestamp(
opts.time_spec)
log.info('Parsed time spec %r as %s (delta: %s)', opts.time_spec,
ts, naturaltime_diff(ts, ts0))
ts_field = 'date_{}'.format(opts.timestamp_type)
ts_field_check = {
'{}__{}'.format(ts_field, 'gt' if opts.newer else 'lt'): ts
}
log.debug('Timestamp field check: %s', ts_field_check)
posts = models.Post.objects.filter(feed__in=feeds,
**ts_field_check)
if log.isEnabledFor(logging.INFO):
if log.isEnabledFor(logging.DEBUG):
log.debug('List of selected Posts (%s):', posts.count())
for post in posts.order_by('pk'):
log.debug(' - [%s] %s', post.pk, post)
else:
log.info(
'Selected %s Post object(s) for'
' cleanup (listed with more verbose logging)',
posts.count())
if not opts.dry_run:
show_count = log.isEnabledFor(logging.INFO) and posts.count()
if show_count:
log.info('Removing %s entries', show_count)
ts0 = time.time()
posts.delete()
if show_count:
log.info('Finished removal of %s entries (time: %.2fs)',
show_count,
time.time() - ts0)
else:
raise ValueError(post.type)
