fediverse/fetchservers.py

464 líneas
11 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
import requests
import urllib3
2021-05-14 10:39:26 +02:00
import socket
import ray
2021-05-14 10:39:26 +02:00
ray.init(num_cpus = 32) # Specify this system CPUs.
2021-05-14 10:39:26 +02:00
2022-05-06 13:53:53 +02:00
from ray.exceptions import (
RaySystemError,
RayError,
RayTaskError,
ObjectStoreFullError,
)
apis = ['/api/v1/instance?',
'/api/v1/nodeinfo?',
'/nodeinfo/2.0?',
'/nodeinfo/2.0.json?',
'/nodeinfo/2.1.json?',
'/main/nodeinfo/2.0?',
'/api/statusnet/config?',
'/api/nodeinfo/2.0.json?',
'/api/nodeinfo?',
'/wp-json/nodeinfo/2.0?',
'/api/v1/instance/nodeinfo/2.0?',
'/.well-known/x-nodeinfo2?'
]
2021-05-14 10:39:26 +02:00
def is_json(myjson):
try:
json_object = json.loads(myjson)
except ValueError as e:
return False
return True
def write_api(server, software, users, alive, api, soft_version):
2021-05-14 10:39:26 +02:00
2022-03-02 22:26:05 +01:00
fediverse_db, fediverse_db_user = get_db_config()
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"
2021-05-14 10:39:26 +02:00
conn = None
try:
2022-03-02 22:26:05 +01:00
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()
2022-03-02 22:26:05 +01:00
print(f'Writing {server} nodeinfo data...')
2021-05-14 10:39:26 +02:00
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:
2022-03-02 22:26:05 +01:00
2021-05-14 10:39:26 +02:00
conn.close()
@ray.remote
def getsoft(server):
if server.find(".") == -1:
return
if server.find("@") != -1:
return
if server.find("/") != -1:
return
if server.find(":") != -1:
return
2022-05-06 13:53:53 +02:00
if server == 'z.fedipen.xyz':
return
soft = ''
is_nodeinfo = False
url = 'https://' + server
user_agent = {'User-agent': "fediverse's stats (fediverse@mastodont.cat)"}
2022-03-02 22:26:05 +01:00
2021-05-14 10:39:26 +02:00
try:
response = requests.get(url + '/.well-known/nodeinfo', headers = user_agent, timeout=3)
2021-05-14 10:39:26 +02:00
if response.status_code == 200:
2021-05-14 10:39:26 +02:00
try:
2021-05-14 10:39:26 +02:00
response_json = response.json()
2021-05-14 10:39:26 +02:00
nodeinfo = response_json['links'][0]['href'].replace(f'https://{server}','')
2021-05-14 10:39:26 +02:00
try:
nodeinfo_data = requests.get(url + nodeinfo, headers = user_agent, timeout=3)
2022-03-02 22:26:05 +01:00
if nodeinfo_data.status_code == 200:
nodeinfo_json = nodeinfo_data.json()
is_nodeinfo = True
else:
print(f"Server {server}'s nodeinfo not responding: error code {nodeinfo_data.status_code}")
except:
2022-03-02 22:26:05 +01:00
pass
2022-03-02 22:26:05 +01:00
except:
2022-03-02 22:26:05 +01:00
print(f'Server {server} not responding: error code {response.status_code}')
print('*********************************************************************')
pass
else:
for api in apis:
try:
response = requests.get(url + api, headers = user_agent, timeout=3)
if is_json(response.text):
nodeinfo_json = response.json()
if 'software' in nodeinfo_json:
nodeinfo = api
is_nodeinfo = True
break
elif 'title' in nodeinfo_json:
if nodeinfo_json['title'] == 'Zap':
nodeinfo = api
is_nodeinfo = True
soft = 'zap'
break
elif 'version' in nodeinfo_json:
nodeinfo = api
is_nodeinfo = True
break
except:
2022-03-02 22:26:05 +01:00
pass
except requests.exceptions.SSLError as errssl:
2022-03-02 22:26:05 +01:00
pass
except requests.exceptions.HTTPError as errh:
2022-03-02 22:26:05 +01:00
pass
2022-03-02 22:26:05 +01:00
except requests.exceptions.ConnectionError as errc:
2022-03-02 22:26:05 +01:00
pass
2022-03-02 22:26:05 +01:00
except requests.exceptions.ReadTimeout as to_err:
pass
2022-03-02 22:26:05 +01:00
except requests.exceptions.TooManyRedirects as tmr_err:
pass
2022-03-02 22:26:05 +01:00
except urllib3.exceptions.LocationParseError as lp_err:
2022-03-02 22:26:05 +01:00
pass
except requests.exceptions.InvalidURL as iu_err:
2022-03-02 22:26:05 +01:00
pass
2022-03-02 22:26:05 +01:00
2022-05-06 13:53:53 +02:00
except requests.exceptions.ChunkedEncodingError as chunk_err:
print(f'ChunkedEncodingError! {server}')
pass
except ray.exceptions.RaySystemError as ray_sys_error:
print(ray_sys_error)
pass
else:
2022-03-02 22:26:05 +01:00
if is_nodeinfo:
2022-03-02 22:26:05 +01:00
if nodeinfo != '/api/v1/instance?':
2022-03-02 22:26:05 +01:00
if nodeinfo != '/.well-known/x-nodeinfo2?':
try:
soft = nodeinfo_json['software']['name']
soft = soft.lower()
soft_version = nodeinfo_json['software']['version']
users = nodeinfo_json['usage']['users']['total']
if users > 1000000:
return
2022-03-02 22:26:05 +01:00
alive = True
write_api(server, soft, users, alive, nodeinfo, soft_version)
print(f"Server {server} ({soft} {soft_version}) is alive!")
2022-03-02 22:26:05 +01:00
print('*********************************************************************')
2021-05-14 13:44:19 +02:00
return
2022-03-02 22:26:05 +01:00
except:
pass
else:
try:
soft = nodeinfo_json['server']['software']
soft = soft.lower()
soft_version = nodeinfo_json['server']['version']
users = nodeinfo_json['usage']['users']['total']
if users > 1000000:
return
alive = True
if soft == 'socialhome':
write_api(server, soft, users, alive, nodeinfo, soft_version)
2022-03-02 22:26:05 +01:00
print('*********************************************************************')
print(f"Server {serve}r ({soft} {soft_version}) is alive!")
2022-03-02 22:26:05 +01:00
print('*********************************************************************')
return
except:
pass
if soft == '' and nodeinfo == "/api/v1/instance?":
2022-03-02 22:26:05 +01:00
soft = 'mastodon'
try:
2022-03-02 22:26:05 +01:00
users = nodeinfo_json['stats']['user_count']
if users > 1000000:
2022-03-02 22:26:05 +01:00
return
2022-03-02 22:26:05 +01:00
except:
2022-03-02 22:26:05 +01:00
users = 0
2022-03-02 22:26:05 +01:00
try:
2021-05-14 10:39:26 +02:00
soft_version = nodeinfo_json['version']
except:
2021-05-14 10:39:26 +02:00
soft_version = 'unknown'
2021-05-14 10:39:26 +02:00
alive = True
2021-05-14 10:39:26 +02:00
write_api(server, soft, users, alive, nodeinfo, soft_version)
2021-05-14 10:39:26 +02:00
print('*********************************************************************')
print(f"Server {server} ({soft}) is alive!")
elif soft == 'zap' and nodeinfo == "/api/v1/instance?":
soft = 'zap'
users = nodeinfo_json['stats']['user_count']
soft_version = nodeinfo_json['version']
alive = True
print(server, soft, users, alive, api)
print('*********************************************************************')
print(f"Server {server} ({soft}) is alive!")
else:
print(f'Server {server} is dead')
print('*********************************************************************')
2021-05-14 10:39:26 +02:00
2022-03-02 22:26:05 +01:00
def get_world_servers():
world_servers = []
try:
conn = None
conn = psycopg2.connect(database=fediverse_db, user=fediverse_db_user, password="", host="/var/run/postgresql", port="5432")
cur = conn.cursor()
# get world servers list
2022-03-24 13:36:59 +01:00
cur.execute("select server from world where checked='f'")
2022-03-02 22:26:05 +01:00
rows = cur.fetchall()
for row in rows:
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()
return world_servers
def save_time(program, start, finish):
insert_sql = "INSERT INTO execution_time(program, start, finish) VALUES(%s,%s,%s) ON CONFLICT DO NOTHING"
conn = None
try:
conn = psycopg2.connect(database = fediverse_db, user = fediverse_db_user, password = "", host = "/var/run/postgresql", port = "5432")
cur = conn.cursor()
cur.execute(insert_sql, (program, start, finish,))
cur.execute("UPDATE execution_time SET start=(%s), finish=(%s) where program=(%s)", (start, finish, program))
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
def get_parameter(parameter, file_path):
2022-03-02 22:26:05 +01:00
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)
def get_config():
# Load configuration from config file
config_filepath = "config/config.txt"
mastodon_hostname = get_parameter("mastodon_hostname", config_filepath)
return mastodon_hostname
2021-05-14 10:39:26 +02:00
def get_db_config():
# 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)
return (fediverse_db, fediverse_db_user)
2021-05-14 10:39:26 +02:00
###############################################################################
# main
if __name__ == '__main__':
2021-05-14 13:44:19 +02:00
## name: fetchservers.py
2021-05-14 13:44:19 +02:00
fediverse_db, fediverse_db_user = get_db_config()
2022-03-02 22:26:05 +01:00
start = datetime.now()
2021-05-14 10:39:26 +02:00
program = 'fetchservers'
finish = start
save_time(program, start, finish)
now = start
mastodon_hostname = get_config()
2021-05-14 10:39:26 +02:00
world_servers = get_world_servers()
2021-05-14 10:39:26 +02:00
ray_start = time.time()
2021-05-14 10:39:26 +02:00
2022-05-06 13:53:53 +02:00
try:
results = ray.get([getsoft.remote(server) for server in world_servers])
2021-05-14 10:39:26 +02:00
2022-05-06 13:53:53 +02:00
print(f"duration = {time.time() - ray_start}.\nprocessed servers: {len(results)}")
except:
pass
finish = datetime.now()
save_time(program, start, finish)