fediverse/fetchservers.py

226 líneas
6.1 KiB
Python
Original Vista normal Històric

2021-05-14 10:39:26 +02:00
import time
from datetime import datetime
import os
import json
import sys
import os.path
import psycopg2
2021-05-14 13:44:19 +02:00
from multiprocessing import Pool, Manager
2021-05-14 10:39:26 +02:00
import aiohttp
import asyncio
import socket
2021-05-14 13:44:19 +02:00
apis = ['/nodeinfo/2.0?', '/nodeinfo/2.0.json?', '/main/nodeinfo/2.0?', '/api/statusnet/config?',
'/api/nodeinfo/2.0.json?', '/api/nodeinfo?', '/api/v1/instance?', '/wp-json/nodeinfo/2.0?']
2021-05-14 10:39:26 +02:00
client_exceptions = (
aiohttp.ClientResponseError,
aiohttp.ClientConnectionError,
aiohttp.ClientConnectorError,
aiohttp.ClientError,
asyncio.TimeoutError,
socket.gaierror,
)
def is_json(myjson):
try:
json_object = json.loads(myjson)
except ValueError as e:
return False
return True
2021-05-14 13:44:19 +02:00
def write_api(server, software, users, alive, api, soft_version):
2021-05-14 10:39:26 +02:00
insert_sql = "INSERT INTO fediverse(server, updated_at, software, users, alive, users_api, version) VALUES(%s,%s,%s,%s,%s,%s,%s) ON CONFLICT DO NOTHING"
conn = None
try:
2021-05-14 13:44:19 +02:00
conn = psycopg2.connect(database=fediverse_db, user=fediverse_db_user, password="", host="/var/run/postgresql",
port="5432")
2021-05-14 10:39:26 +02:00
cur = conn.cursor()
cur.execute(insert_sql, (server, now, software, users, alive, api, soft_version))
2021-05-14 13:44:19 +02:00
cur.execute(
"UPDATE fediverse SET updated_at=(%s), software=(%s), users=(%s), alive=(%s), users_api=(%s), version=(%s) where server=(%s)",
(now, software, users, alive, api, soft_version, server))
2021-05-14 10:39:26 +02:00
cur.execute("UPDATE world SET checked='t' where server=(%s)", (server,))
conn.commit()
cur.close()
except (Exception, psycopg2.DatabaseError) as error:
print(error)
finally:
if conn is not None:
conn.close()
2021-05-14 13:44:19 +02:00
async def getsoft(server):
2021-05-29 12:44:12 +02:00
2021-05-14 10:39:26 +02:00
try:
socket.gethostbyname(server)
except socket.gaierror:
pass
return
soft = ''
url = 'https://' + server
timeout = aiohttp.ClientTimeout(total=3)
async with aiohttp.ClientSession(timeout=timeout) as session:
for api in apis:
try:
2021-05-14 13:44:19 +02:00
async with session.get(url + api) as response:
2021-05-14 10:39:26 +02:00
if response.status == 200:
try:
response_json = await response.json()
except:
pass
except aiohttp.ClientConnectorError as err:
pass
else:
2021-05-14 13:44:19 +02:00
if response.status == 200 and api != '/api/v1/instance?':
try:
soft = response_json['software']['name']
soft = soft.lower()
soft_version = response_json['software']['version']
users = response_json['usage']['users']['total']
if users > 1000000:
return
alive = True
write_api(server, soft, users, alive, api, soft_version)
print("Server " + server + " (" + soft + " " + soft_version + ") is alive!")
return
except:
pass
if response.status == 200 and soft == '' and api == "/api/v1/instance?":
soft = 'mastodon'
users = response_json['stats']['user_count']
soft_version = response_json['version']
if users > 1000000:
return
alive = True
write_api(server, soft, users, alive, api)
print("Server " + server + " (" + soft + ") is alive!")
2021-05-14 10:39:26 +02:00
def getserver(server, x):
server = server[0].rstrip('.').lower()
2021-05-29 12:44:12 +02:00
2021-05-14 10:39:26 +02:00
if server.find(".") == -1:
return
if server.find("@") != -1:
2021-05-14 13:44:19 +02:00
return
2021-05-14 10:39:26 +02:00
if server.find("/") != -1:
2021-05-14 13:44:19 +02:00
return
2021-05-14 10:39:26 +02:00
if server.find(":") != -1:
2021-05-14 13:44:19 +02:00
return
2021-05-14 10:39:26 +02:00
try:
loop = asyncio.get_event_loop()
coroutines = [getsoft(server)]
soft = loop.run_until_complete(asyncio.gather(*coroutines, return_exceptions=True))
except:
pass
2021-05-14 13:44:19 +02:00
2021-05-14 10:39:26 +02:00
# Returns the parameter from the specified file
2021-05-14 13:44:19 +02:00
def get_parameter(parameter, file_path):
2021-05-14 10:39:26 +02:00
# Check if secrets file exists
if not os.path.isfile(file_path):
2021-05-14 13:44:19 +02:00
print("File %s not found, exiting." % file_path)
2021-05-14 10:39:26 +02:00
sys.exit(0)
# Find parameter in file
2021-05-14 13:44:19 +02:00
with open(file_path) as f:
2021-05-14 10:39:26 +02:00
for line in f:
2021-05-14 13:44:19 +02:00
if line.startswith(parameter):
2021-05-14 10:39:26 +02:00
return line.replace(parameter + ":", "").strip()
# Cannot find parameter, exit
2021-05-14 13:44:19 +02:00
print(file_path + " Missing parameter %s " % parameter)
2021-05-14 10:39:26 +02:00
sys.exit(0)
# Load configuration from config file
config_filepath = "config/config.txt"
mastodon_hostname = get_parameter("mastodon_hostname", config_filepath)
# Load database config from db_config file
db_config_filepath = "config/db_config.txt"
fediverse_db = get_parameter("fediverse_db", db_config_filepath)
fediverse_db_user = get_parameter("fediverse_db_user", db_config_filepath)
###############################################################################
# main
if __name__ == '__main__':
2021-05-29 12:44:12 +02:00
2021-05-14 13:44:19 +02:00
now = datetime.now()
start_time = time.time()
2021-05-14 10:39:26 +02:00
world_servers = []
try:
conn = None
2021-05-14 13:44:19 +02:00
2021-05-29 12:44:12 +02:00
conn = psycopg2.connect(database=fediverse_db, user=fediverse_db_user, password="", host="/var/run/postgresql", port="5432")
2021-05-14 10:39:26 +02:00
cur = conn.cursor()
2021-05-14 13:44:19 +02:00
# get world servers list
2021-05-14 10:39:26 +02:00
cur.execute("select server from world where checked='f'")
2021-05-29 12:44:12 +02:00
rows = cur.fetchall()
for row in rows:
2021-05-14 10:39:26 +02:00
world_servers.append(row[0])
cur.close()
print("Remaining servers: " + str(len(world_servers)))
except (Exception, psycopg2.DatabaseError) as error:
print(error)
finally:
if conn is not None:
conn.close()
###########################################################################
# multiprocessing!
m = Manager()
q = m.Queue()
z = zip(world_servers)
serv_number = len(world_servers)
pool_tuple = [(x, q) for x in z]
with Pool(processes=64) as pool:
pool.starmap(getserver, pool_tuple)
print('Done.')