def test_simple_tarball_dump(named_temporary_file):
voevent_etrees = fake.heartbeat_packets()
voevent_rowgen = (models.Voevent.from_etree(v) for v in voevent_etrees)
fname = named_temporary_file.name
filestore.write_tarball(voevent_rowgen, fname)
tarball = tarfile.open(fname)
assert len(tarball.getmembers()) == len(voevent_etrees)
def test_tarball_round_trip(named_temporary_file, fixture_db_session):
voevent_etrees = fake.heartbeat_packets()
# with open(assasn_non_ascii_packet_filepath, 'rb') as f:
# voevent_etrees.append(vp.load(f))
s = fixture_db_session
for etree in voevent_etrees:
s.add(models.Voevent.from_etree(etree))
s.flush()
voevent_dbrows = s.query(models.Voevent.ivorn, models.Voevent.xml).all()
assert len(voevent_dbrows) == len(voevent_etrees)
voevent_rowgen = list(models.Voevent.from_etree(v) for v in voevent_etrees)
assert voevent_dbrows[0].ivorn == voevent_rowgen[0].ivorn
assert voevent_dbrows[0].xml == voevent_rowgen[0].xml
assert type(voevent_dbrows[0].xml) == type(voevent_rowgen[0].xml)
assert type(voevent_rowgen[0].xml) == six.binary_type
# Therefore it's crucial to test with an actual round-tripped dataset,
# the 'voevent_dbrows' from above:
fname = named_temporary_file.name
filestore.write_tarball(voevent_dbrows, fname)
loaded_voevents = [
vp.loads(s.xml) for s in filestore.tarfile_xml_generator(fname)
]
def to_strings(voeventlist):
return [vp.dumps(v) for v in voeventlist]
def to_ivorn(voeventlist):
return [v.attrib['ivorn'] for v in voeventlist]
assert (to_ivorn(voevent_etrees) == to_ivorn(loaded_voevents))
assert (to_strings(voevent_etrees) == to_strings(loaded_voevents))
def test_simple_tarball_dump(named_temporary_file):
voevent_etrees = fake.heartbeat_packets()
voevent_rowgen = (models.Voevent.from_etree(v) for v in voevent_etrees)
fname = named_temporary_file.name
filestore.write_tarball(voevent_rowgen, fname)
tarball = tarfile.open(fname)
assert len(tarball.getmembers()) == len(voevent_etrees)
def test_unicode_voevent_tarball_dump(named_temporary_file):
## Now try some unicode characters
voevent_etrees = fake.heartbeat_packets()
vp.set_author(voevent_etrees[0], contactName=u"€€€€")
voevent_rowgen = (models.Voevent.from_etree(v) for v in voevent_etrees)
fname = named_temporary_file.name
filestore.write_tarball(voevent_rowgen, fname)
tarball = tarfile.open(fname)
assert len(tarball.getmembers()) == len(voevent_etrees)
def test_unicode_voevent_tarball_dump(named_temporary_file):
## Now try some unicode characters
voevent_etrees = fake.heartbeat_packets()
vp.set_author(voevent_etrees[0], contactName=u"€€€€")
voevent_rowgen = (models.Voevent.from_etree(v) for v in voevent_etrees)
fname = named_temporary_file.name
filestore.write_tarball(voevent_rowgen, fname)
tarball = tarfile.open(fname)
assert len(tarball.getmembers()) == len(voevent_etrees)
def main():
args = handle_args()
dburl = dbconfig.make_db_url(dbconfig.default_admin_db_params, args.dbname)
if not db_utils.check_database_exists(dburl):
raise RuntimeError("Database not found")
filecount = 1
n_packets_written = 0
def get_tarfile_path():
if args.nsplit:
suffix = '.{0:03d}.tar.bz2'.format(filecount)
else:
suffix = '.tar.bz2'
return args.tarfile_pathstem + suffix
session = Session(bind=create_engine(dburl))
if args.prefetch:
qry = session.query(Voevent.ivorn, Voevent.xml)
else:
qry = session.query(Voevent)
if args.all:
logger.info("Dumping **all** packets currently in database")
else:
qry = qry.filter(Voevent.author_datetime < args.end)
if args.start is not None:
qry = qry.filter(Voevent.author_datetime >= args.start)
logger.info("Fetching packets from {}".format(args.start))
else:
logger.info("Fetching packets from beginning of time")
logger.info("...until: {}".format(args.end))
qry = qry.order_by(Voevent.id)
n_matching = qry.count()
logger.info("Dumping {} packets".format(n_matching))
start_time = datetime.datetime.now()
while n_packets_written < n_matching:
logger.debug("Fetching batch of up to {} packets".format(args.nsplit))
voevents = qry.limit(args.nsplit).offset(n_packets_written).all()
n_packets_written += write_tarball(voevents,
get_tarfile_path())
elapsed = (datetime.datetime.now() - start_time).total_seconds()
logger.info(
"{} packets dumped so far, in {} ({:.0f} kilopacket/s)".format(
n_packets_written,
elapsed,
n_packets_written / elapsed
))
filecount += 1
session.close()
logger.info("Wrote {} packets".format(n_packets_written))
return 0
def test_tarball_round_trip(named_temporary_file, fixture_db_session):
voevent_etrees = fake.heartbeat_packets()
# with open(assasn_non_ascii_packet_filepath, 'rb') as f:
# voevent_etrees.append(vp.load(f))
s = fixture_db_session
for etree in voevent_etrees:
s.add(models.Voevent.from_etree(etree))
s.flush()
voevent_dbrows = s.query(models.Voevent.ivorn, models.Voevent.xml).all()
assert len(voevent_dbrows) == len(voevent_etrees)
voevent_rowgen = list(models.Voevent.from_etree(v) for v in voevent_etrees)
assert voevent_dbrows[0].ivorn == voevent_rowgen[0].ivorn
assert voevent_dbrows[0].xml == voevent_rowgen[0].xml
# Here's the crux
# A newly instantiated model will store a string type same as Python 2;
# bytestring stores as bytestring, unicode as unicode.
# However, after a round-trip to the database, proper typing has been
# asserted and the bytestring is returned as unicode!
assert type(voevent_dbrows[0].xml) != type(voevent_rowgen[0].xml)
assert type(voevent_rowgen[0].xml) == str
assert type(voevent_dbrows[0].xml) == unicode
# Therefore it's crucial to test with an actual round-tripped dataset,
# the 'voevent_dbrows' from above:
fname = named_temporary_file.name
filestore.write_tarball(voevent_dbrows, fname)
loaded_voevents = [vp.loads(s.xml) for s in
filestore.tarfile_xml_generator(fname)]
def to_strings(voeventlist):
return [vp.dumps(v) for v in voeventlist]
def to_ivorn(voeventlist):
return [v.attrib['ivorn'] for v in voeventlist]
assert (to_ivorn(voevent_etrees) == to_ivorn(loaded_voevents))
assert (to_strings(voevent_etrees) == to_strings(loaded_voevents))
def main():
args = handle_args()
dburl = dbconfig.make_db_url(dbconfig.default_admin_db_params, args.dbname)
if not db_utils.check_database_exists(dburl):
raise RuntimeError("Database not found")
filecount = 1
n_packets_written = 0
def get_tarfile_path():
if args.nsplit:
suffix = '.{0:03d}.tar.bz2'.format(filecount)
else:
suffix = '.tar.bz2'
return args.tarfile_pathstem + suffix
session = Session(bind=create_engine(dburl))
if args.prefetch:
qry = session.query(Voevent.ivorn, Voevent.xml)
else:
qry = session.query(Voevent)
if args.all:
logger.info("Dumping **all** packets currently in database")
else:
qry = qry.filter(Voevent.author_datetime < args.end)
if args.start is not None:
qry = qry.filter(Voevent.author_datetime >= args.start)
logger.info("Fetching packets from {}".format(args.start))
else:
logger.info("Fetching packets from beginning of time")
logger.info("...until: {}".format(args.end))
qry = qry.order_by(Voevent.id)
n_matching = qry.count()
logger.info("Dumping {} packets".format(n_matching))
start_time = datetime.datetime.now()
while n_packets_written < n_matching:
logger.debug("Fetching batch of up to {} packets".format(args.nsplit))
voevents = qry.limit(args.nsplit).offset(n_packets_written).all()
n_packets_written += write_tarball(voevents, get_tarfile_path())
elapsed = (datetime.datetime.now() - start_time).total_seconds()
logger.info(
"{} packets dumped so far, in {} ({:.0f} kilopacket/s)".format(
n_packets_written, elapsed, n_packets_written / elapsed))
filecount += 1
session.close()
logger.info("Wrote {} packets".format(n_packets_written))
return 0