def parse_ips(jobs: Queue):
session = setup_connection()
counter = 0
start_time = time.time()
while True:
block = jobs.get()
if block is None:
break
ip_start = block.inetnum.split(" - ")[0]
ip_end = block.inetnum.split(" - ")[1]
cidrs = iprange_to_cidrs(ip_start, ip_end)
for c in cidrs:
session.add(Cidr(cidr=str(c), block=block))
counter += 1
if counter % COMMIT_COUNT == 0:
session.commit()
session.close()
session = setup_connection()
logger.debug("committed cidrs ({} seconds)".format(round(time.time() - start_time, 2)))
counter = 0
start_time = time.time()
session.commit()
logger.debug("committed last cidrs")
session.close()
logger.debug("{} finished".format(current_process().name))
def parse_blocks(jobs: Queue, connection_string: str):
session = setup_connection(connection_string)
counter = 0
BLOCKS_DONE = 0
start_time = time.time()
while True:
block = jobs.get()
if block is None:
break
inetnum = parse_property_inetnum(block)
netname = parse_property(block, 'netname')
description = parse_property(block, 'descr')
country = parse_property(block, 'country')
maintained_by = parse_property(block, 'mnt-by')
created = parse_property(block, 'created')
last_modified = parse_property(block, 'last-modified')
source = parse_property(block, 'cust_source')
b = Block(inetnum=inetnum,
netname=netname,
description=description,
country=country,
maintained_by=maintained_by,
created=created,
last_modified=last_modified,
source=source)
session.add(b)
counter += 1
BLOCKS_DONE += 1
if counter % COMMIT_COUNT == 0:
session.commit()
session.close()
session = setup_connection(connection_string)
# not really accurate at the moment
percent = (BLOCKS_DONE * NUM_WORKERS * 100) / NUM_BLOCKS
if percent > 100:
percent = 100
logger.debug(
'committed {} blocks ({} seconds) {:.1f}% done.'.format(
counter, round(time.time() - start_time, 2), percent))
counter = 0
start_time = time.time()
session.commit()
logger.debug('committed last blocks')
session.close()
logger.debug('{} finished'.format(current_process().name))
def main(connection_string):
overall_start_time = time.time()
session = setup_connection(connection_string, create_db=True)
for entry in FILELIST:
global CURRENT_FILENAME
CURRENT_FILENAME = entry
f_name = f"./databases/{entry}"
if os.path.exists(f_name):
logger.info(f"parsing database file: {f_name}")
start_time = time.time()
blocks = read_blocks(f_name)
logger.info(
f"database parsing finished: {round(time.time() - start_time, 2)} seconds"
)
logger.info('parsing blocks')
start_time = time.time()
jobs = Queue()
workers = []
# start workers
logger.debug(f"starting {NUM_WORKERS} processes")
for w in range(NUM_WORKERS):
p = Process(target=parse_blocks,
args=(
jobs,
connection_string,
),
daemon=True)
p.start()
workers.append(p)
# add tasks
for b in blocks:
jobs.put(b)
for i in range(NUM_WORKERS):
jobs.put(None)
jobs.close()
jobs.join_thread()
# wait to finish
for p in workers:
p.join()
logger.info(
f"block parsing finished: {round(time.time() - start_time, 2)} seconds"
)
else:
logger.info(
f"File {f_name} not found. Please download using download_dumps.sh"
)
CURRENT_FILENAME = "empty"
logger.info(
f"script finished: {round(time.time() - overall_start_time, 2)} seconds"
)
def main(connection_string):
overall_start_time = time.time()
session = setup_connection(connection_string, create_db=True)
for entry in FILELIST:
global CURRENT_FILENAME
CURRENT_FILENAME = entry
f_name = "./databases/{}".format(entry)
if os.path.exists(f_name):
logger.info('parsing database file: {}'.format(f_name))
start_time = time.time()
blocks = read_blocks(f_name)
logger.info('database parsing finished: {} seconds'.format(
round(time.time() - start_time, 2)))
logger.info('parsing blocks')
start_time = time.time()
jobs = Queue()
workers = []
# start workers
logger.debug('starting {} processes'.format(NUM_WORKERS))
for w in range(NUM_WORKERS):
p = Process(target=parse_blocks,
args=(
jobs,
connection_string,
))
p.start()
workers.append(p)
# add tasks
for b in blocks:
jobs.put(b)
for i in range(NUM_WORKERS):
jobs.put(None)
# wait to finish
for p in workers:
p.join()
logger.info('block parsing finished: {} seconds'.format(
round(time.time() - start_time, 2)))
else:
logger.info(
'File {} not found. Please download using download_dumps.sh'.
format(f_name))
CURRENT_FILENAME = "empty"
logger.info('script finished: {} seconds'.format(
round(time.time() - overall_start_time, 2)))
def parse_blocks(jobs: Queue):
session = setup_connection()
counter = 0
start_time = time.time()
while True:
block = jobs.get()
if block is None:
break
inetnum = parse_property(block, "inetnum")
netname = parse_property(block, "netname")
description = parse_property(block, "descr")
country = parse_property(block, "country")
maintained_by = parse_property(block, "mnt-by")
created = parse_property(block, "created")
last_modified = parse_property(block, "last-modified")
b = Block(
inetnum=inetnum,
netname=netname,
description=description,
country=country,
maintained_by=maintained_by,
created=created,
last_modified=last_modified,
)
session.add(b)
counter += 1
if counter % COMMIT_COUNT == 0:
session.commit()
session.close()
session = setup_connection()
logger.debug("committed {} blocks ({} seconds)".format(counter, round(time.time() - start_time, 2)))
counter = 0
start_time = time.time()
session.commit()
logger.debug("committed last blocks")
session.close()
logger.debug("{} finished".format(current_process().name))
def main():
overall_start_time = time.time()
session = setup_connection(create_db=True)
logger.info("parsing ripe database")
start_time = time.time()
blocks = read_blocks()
logger.info("ripe database parsing finished: {} seconds".format(round(time.time() - start_time, 2)))
logger.info("parsing blocks")
start_time = time.time()
jobs = Queue()
workers = []
# start workers
logger.debug("starting {} processes".format(NUM_WORKERS))
for w in range(NUM_WORKERS):
p = Process(target=parse_blocks, args=(jobs,))
p.start()
workers.append(p)
# add tasks
for b in blocks:
jobs.put(b)
for i in range(NUM_WORKERS):
jobs.put(None)
# wait to finish
for p in workers:
p.join()
logger.info("block parsing finished: {} seconds".format(round(time.time() - start_time, 2)))
logger.info("parsing IPs")
start_time = time.time()
jobs = Queue()
workers = []
# start workers
logger.debug("starting {} processes".format(NUM_WORKERS))
for w in range(NUM_WORKERS):
p = Process(target=parse_ips, args=(jobs,))
p.start()
workers.append(p)
# add tasks
logger.debug("populating job queue")
for b in session.query(Block):
jobs.put(b)
session.close()
for i in range(NUM_WORKERS):
jobs.put(None)
logger.debug("job queue populated")
# wait to finish
for p in workers:
p.join()
logger.info("ip parsing finished: {} seconds".format(round(time.time() - start_time, 2)))
logger.info("script finished: {} seconds".format(round(time.time() - overall_start_time, 2)))
