Comparar commits

..

No hay commits en común. '5992e671720651c51031cb2780d269710b04e571' y '081afcde0d3d37801e610b85256a21e40c0e46ba' tienen historias completamente diferentes.

S'han modificat 6 arxius amb 1374 adicions i 425 eliminacions

Veure arxiu

@ -1,5 +1,5 @@
# Fediverse Stats
This code gets all peers from mastodon.social. Goal is to collect maximum number
This code gets all peers from running Mastodon, Pleroma and Lemmy host servers and then all peers from host server's peers. Goal is to collect maximum number
of alive fediverse's servers and then query their API to obtain their registered users (if their API provide such information).
At the end it post the results to host server bot account.
@ -7,7 +7,7 @@ At the end it post the results to host server bot account.
- **Python 3**
- Postgresql server
- Mastodon running server.
- Mastodon or Pleroma running server.
### Usage:
@ -15,11 +15,19 @@ Within Python Virtual Environment:
1. Run `pip install -r requirements.txt` to install needed libraries.
2. Run `python fetchservers.py` to add servers to alive servers database.
2. Run `python db-setup.py` to setup and create new Postgresql database and needed tables in it.
3. Run `python fediverse.py` to query world alive servers API. It gets data from server's nodeinfo.
3. Run `python setup.py` to get your bot's access token of your Mastodon or Pleroma server existing account. It will be saved to 'secrets/secrets.txt' for further use.
4. Use your favourite scheduling method to set `python fediverse.py` to run twice daily, `python fetchservers.py` one time daily.
4. Run `python getpeers.py` to get all peers from your host and the whole world of fediverse's servers (or almost the whole world).
5. Run `python fetchservers.py` to add servers to alive servers database.
6. Run `python fediverse.py` to query world alive servers API. It gets data from server's nodeinfo.
7. Run `python uptime_setup.py` to get your Uptime bot's access token of your Mastodon or Pleroma server existing account. It will be saved to 'secrets/uptime_secrets.txt' for further use.
8. Use your favourite scheduling method to set `python fediverse.py` to run twice daily, `python fetchservers.py` one time daily, `python getworld.py` to run monthly and `python uptime.py` (choose your desired frequency) if you want to publish best fediverse's servers uptime.
18.2.2021 - New feature! Added [Lemmy project](https://join.lemmy.ml)
12.5.2021 - New feature! Added Wordpress support. The code can now detect Wordpress instances with ActivityPub enabled plugin.
@ -27,6 +35,4 @@ Within Python Virtual Environment:
21.8.2021 - New feature! Added Best Fediverse's servers Uptime publishing bot.
22.10.2021 - New feature! Added [Funkwhale](https://funkwhale.audio) support.
26.10.2021 - New feature! Added [Socialhome](https://socialhome.network) support.
2.3.2022 - Improved server nodeinfo detection.
4.1.2023 - Refactored.
4.1.2023 - Now all peers are obtained from mastodon.social's peers list.
2.3.2022 - Improved server nodeinfo detection

Veure arxiu

@ -5,33 +5,13 @@ from datetime import datetime
import urllib3
import requests
import socket
from setup import Setup
from database import Database
from mastodon import Mastodon
from matplotlib import pyplot as plt
import matplotlib.dates as mdates
from matplotlib.ticker import ScalarFormatter
import numpy as np
import pandas as pd
import psycopg2
import matplotlib.pyplot as plt
import ray
import pdb
SMALL_SIZE = 6
MEDIUM_SIZE = 10
BIGGER_SIZE = 12
plt.rc('font', size=MEDIUM_SIZE) # controls default text sizes
plt.rc('axes', titlesize=MEDIUM_SIZE) # fontsize of the axes title
plt.rc('axes', labelsize=MEDIUM_SIZE) # fontsize of the x and y labels
plt.rc('xtick', labelsize=SMALL_SIZE) # fontsize of the tick labels
plt.rc('ytick', labelsize=MEDIUM_SIZE) # fontsize of the tick labels
plt.rc('legend', fontsize=MEDIUM_SIZE) # legend fontsize
plt.rc('figure', titlesize=BIGGER_SIZE) # fontsize of the figure title
mdates.set_epoch('2000-01-01T00:00:00')
y_formatter = ScalarFormatter(useOffset=False)
ray.init(num_cpus = 25) # Specify this system CPUs.
ray.init(num_cpus = 32) # Specify this system CPUs.
class Server:
@ -48,35 +28,76 @@ class Server:
downs = 0
was_alive, software, api, soft_version, first_checked_at, downs_qty = db.get_server_data(self)
try:
conn = None
conn = psycopg2.connect(database=fediverse_db, user=fediverse_db_user, password="", host="/var/run/postgresql", port="5432")
cur = conn.cursor()
cur.execute("select alive, software, users_api, version, first_checked_at, downs from fediverse where server=(%s)", (self,))
row = cur.fetchone()
if row is not None:
was_alive = row[0]
software = row[1]
api = row[2]
soft_version = row[3]
first_checked_at = row[4]
downs_qty = row[5]
cur.close()
except (Exception, psycopg2.DatabaseError) as error:
print(error)
finally:
if conn is not None:
conn.close()
alive = False
try:
data = requests.get('https://' + self + api, headers = setup.user_agent, timeout=3)
user_agent = {'User-agent': "fediverse's servers stats (fediverse@mastodont.cat)"}
nodeinfo_json = data.json()
data = requests.get('https://' + self + api, headers = user_agent, timeout=3)
try:
users = nodeinfo_json.get('usage').get('users').get('total') or '0'
users = data.json()['usage']['users']['total']
mau = nodeinfo_json.get('usage').get('users').get('activeMonth') or '0'
if users == 0:
if software == 'socialhome':
users = data.json()['usage']['users']['activeHalfyear']
soft_version = nodeinfo_json['server']['version']
if 'activeMonth' in data.json()['usage']['users'].keys():
mau = data.json()['usage']['users']['activeMonth']
else:
soft_version = nodeinfo_json['software']['version']
mau = 0
if software == "wordpress" and "activitypub" in nodeinfo_json['protocols']:
if software == 'socialhome':
soft_version = data.json()['server']['version']
else:
soft_version = data.json()['software']['version']
if software == "wordpress" and "activitypub" in data.json()['protocols']:
alive = True
elif software == "wordpress" and "activitypub" not in nodeinfo_json['protocols']:
elif software == "wordpress" and "activitypub" not in data.json()['protocols']:
alive = False
@ -93,8 +114,8 @@ class Server:
if api == '/api/v1/instance?':
try:
users = nodeinfo_json.get('stats').get('user_count') or '0'
soft_version = nodeinfo_json['version']
users = data.json()['stats']['user_count']
soft_version = data.json()['version']
alive = True
mau = 0
except:
@ -116,11 +137,11 @@ class Server:
if software != 'birdsitelive':
db.write_alive_server(self, software, soft_version, alive, api, users, downs, first_checked_at, mau)
write_alive_server(self, software, soft_version, alive, api, users, downs, first_checked_at, mau)
else:
db.write_blocked_software(self, software, soft_version, alive, api, users, downs, first_checked_at)
write_blocked_software(self, software, soft_version, alive, api, users, downs, first_checked_at)
except urllib3.exceptions.ProtocolError as protoerr:
@ -184,19 +205,325 @@ class Server:
downs = 1
db.write_not_alive_server(self, software, soft_version, alive, api, users, downs, first_checked_at)
write_not_alive_server(self, software, soft_version, alive, api, users, downs, first_checked_at)
return (self, software, soft_version, alive, api, users, downs, first_checked_at, mau)
def write_alive_server(server, software, soft_version, alive, api, users, downs, first_checked_at, mau):
insert_sql = "INSERT INTO fediverse(server, users, updated_at, software, alive, users_api, version, first_checked_at, last_checked_at, downs, mau) VALUES(%s,%s,%s,%s,%s,%s,%s,%s,%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, (server, users, now, software, alive, api, soft_version, now, now, downs, mau))
if first_checked_at != None:
cur.execute("UPDATE fediverse SET users=(%s), updated_at=(%s), software=(%s), alive=(%s), users_api=(%s), version=(%s), last_checked_at=(%s), downs=(%s), mau=(%s) where server=(%s)", (users, now, software, alive, api, soft_version, now, downs, mau, server))
else:
cur.execute("UPDATE fediverse SET users=(%s), updated_at=(%s), software=(%s), alive=(%s), users_api=(%s), version=(%s), first_checked_at=(%s), last_checked_at=(%s), downs=(%s), mau=(%s) where server=(%s)", (users, now, software, alive, api, soft_version, now, now, downs, mau, server))
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()
def write_blocked_software(server, software, soft_version, alive, api, users, downs, first_checked_at):
insert_sql = "INSERT INTO blocked(server, users, updated_at, software, alive, users_api, version, first_checked_at, last_checked_at, downs) VALUES(%s,%s,%s,%s,%s,%s,%s,%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, (server, users, now, software, alive, api, soft_version, now, now, downs))
if first_checked_at != None:
cur.execute("UPDATE blocked SET users=(%s), updated_at=(%s), software=(%s), alive=(%s), users_api=(%s), version=(%s), last_checked_at=(%s), downs=(%s) where server=(%s)", (users, now, software, alive, api, soft_version, now, downs, server))
else:
cur.execute("UPDATE blocked SET users=(%s), updated_at=(%s), software=(%s), alive=(%s), users_api=(%s), version=(%s), first_checked_at=(%s), last_checked_at=(%s), downs=(%s) where server=(%s)", (users, now, software, alive, api, soft_version, now, now, downs, server))
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()
def write_not_alive_server(server, software, soft_version, alive, api, users, downs, first_checked_at):
conn = None
try:
conn = psycopg2.connect(database=fediverse_db, user=fediverse_db_user, password="", host="/var/run/postgresql", port="5432")
cur = conn.cursor()
if first_checked_at != None:
cur.execute("UPDATE fediverse SET updated_at=(%s), alive=(%s), downs=(%s) where server=(%s)", (now, alive, downs, server))
else:
cur.execute("UPDATE fediverse SET updated_at=(%s), alive=(%s), first_checked_at=(%s), downs=(%s) where server=(%s)", (now, alive, now, downs, server))
cur.execute("UPDATE world SET checked='f' where server=(%s)", (server,))
conn.commit()
cur.close()
except (Exception, psycopg2.DatabaseError) as error:
print(error)
finally:
if conn is not None:
conn.close()
def delete_dead_servers():
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("select server from fediverse where downs > '14' and not alive and now() - first_checked_at > interval '7 days'")
rows = cur.fetchall()
for row in rows:
print(f'Deleting server {row[0]}...')
cur.execute("delete from fediverse where server=(%s)", (row[0],))
conn.commit()
cur.close()
except (Exception, psycopg2.DatabaseError) as error:
print(error)
finally:
if conn is not None:
conn.close()
def get_last_checked_servers():
############################################################################
# get last checked servers from fediverse DB
alive_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
cur.execute("select server from world where server in (select server from fediverse where users_api != '')")
alive_servers = []
for row in cur:
alive_servers.append(row[0])
cur.close()
except (Exception, psycopg2.DatabaseError) as error:
print(error)
finally:
if conn is not None:
conn.close()
return alive_servers
def print_dead(server):
print(f'\nServer {server} is dead :-(')
def set_world_servers_check_to_false():
############################################################################
# set all world servers's checked column to False
try:
conn = None
conn = psycopg2.connect(database=fediverse_db, user=fediverse_db_user, password="", host="/var/run/postgresql", port="5432")
cur = conn.cursor()
cur.execute("UPDATE world SET checked='f'")
conn.commit()
cur.close()
except (Exception, psycopg2.DatabaseError) as error:
print(error)
finally:
if conn is not None:
conn.close()
def mastodon():
# Load secrets from secrets file
secrets_filepath = "secrets/secrets.txt"
uc_client_id = get_parameter("uc_client_id", secrets_filepath)
uc_client_secret = get_parameter("uc_client_secret", secrets_filepath)
uc_access_token = get_parameter("uc_access_token", secrets_filepath)
# Load configuration from config file
config_filepath = "config/config.txt"
mastodon_hostname = get_parameter("mastodon_hostname", config_filepath)
# Initialise Mastodon API
mastodon = Mastodon(
client_id=uc_client_id,
client_secret=uc_client_secret,
access_token=uc_access_token,
api_base_url='https://' + mastodon_hostname,
)
# Initialise access headers
headers = {'Authorization': 'Bearer %s'%uc_access_token}
return (mastodon, mastodon_hostname)
def db_config():
# Load db configuration from config file
config_filepath = "config/db_config.txt"
fediverse_db = get_parameter("fediverse_db", config_filepath)
fediverse_db_user = get_parameter("fediverse_db_user", config_filepath)
return (fediverse_db, fediverse_db_user)
def usage():
print('usage: python ' + sys.argv[0] + ' --multi' + ' (multiprocessing, fast)')
print('usage: python ' + sys.argv[0] + ' --mono' + ' (one process, slow)')
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()
def get_parameter(parameter, file_path):
# Check if secrets file exists
if not os.path.isfile(file_path):
print("File %s not found, exiting."%file_path)
sys.exit(0)
# Find parameter in file
with open(file_path) as f:
for line in f:
if line.startswith(parameter):
return line.replace(parameter + ":", "").strip()
# Cannot find parameter, exit
print(file_path + " Missing parameter %s "%parameter)
sys.exit(0)
if __name__ == '__main__':
db = Database()
# usage modes
setup = Setup()
if len(sys.argv) == 1:
usage()
elif len(sys.argv) == 2:
if sys.argv[1] == '--multi':
fediverse_db, fediverse_db_user = db_config()
start = datetime.now()
@ -204,20 +531,19 @@ if __name__ == '__main__':
finish = start
db.save_time(program, start, finish)
save_time(program, start, finish)
now = start
mastodon = Mastodon(
access_token = setup.mastodon_app_token,
api_base_url= setup.mastodon_hostname
)
mastodon, mastodon_hostname = mastodon()
total_servers = 0
total_users = 0
alive_servers = db.get_last_checked_servers()
#set_world_servers_check_to_false()
alive_servers = get_last_checked_servers()
getservers = Server()
@ -229,29 +555,276 @@ if __name__ == '__main__':
print(f"duration = {time.time() - ray_start}.\nprocessed servers: {len(results)}")
###########################################################################
# get current total servers and users, get users from every software
soft_total_project, soft_total_users, soft_total_mau, soft_total_servers, total_servers, total_users, total_mau = db.soft_totals()
now = datetime.now()
gettotals_sql = "select count(server), sum(users), sum(mau) from fediverse where alive"
get_soft_totals_sql = "select software, sum(users) as users, sum(mau) as mau, count(server) as servers from fediverse where users != 0 and mau is not null and alive group by software order by mau desc"
soft_total_project = []
soft_total_users = []
soft_total_mau = []
soft_total_servers = []
try:
conn = None
conn = psycopg2.connect(database=fediverse_db, user=fediverse_db_user, password="", host="/var/run/postgresql", port="5432")
cur = conn.cursor()
cur.execute(gettotals_sql)
row = cur.fetchone()
total_servers = row[0]
total_users = row[1]
total_mau = row[2]
cur.execute(get_soft_totals_sql)
rows = cur.fetchall()
for row in rows:
soft_total_project.append(row[0])
soft_total_users.append(row[1])
soft_total_mau.append(row[2])
soft_total_servers.append(row[3])
cur.close()
except (Exception, psycopg2.DatabaseError) as error:
print(error)
finally:
if conn is not None:
conn.close()
###########################################################################
# get last check values and write current total ones
evo_servers, evo_users, evo_mau = db.last_values(total_servers, total_users, total_mau)
select_sql = "select total_servers, total_users, total_mau from totals order by datetime desc limit 1"
insert_sql = "INSERT INTO totals(datetime, total_servers, total_users, total_mau) VALUES(%s,%s,%s,%s)"
try:
conn = None
conn = psycopg2.connect(database=fediverse_db, user=fediverse_db_user, password="", host="/var/run/postgresql", port="5432")
cur = conn.cursor()
cur.execute(select_sql)
row = cur.fetchone()
if row is not None:
servers_before = row[0]
users_before = row[1]
mau_before = row[2]
else:
servers_before = 0
users_before = 0
mau_before = 0
cur.execute(insert_sql, (now, total_servers, total_users, total_mau))
conn.commit()
cur.close()
evo_servers = total_servers - servers_before
evo_users = total_users - users_before
evo_mau = total_mau - mau_before
except (Exception, psycopg2.DatabaseError) as error:
print(error)
finally:
if conn is not None:
conn.close()
################################################################################
# write evo values
db.write_evo(evo_servers, evo_users, evo_mau)
insert_sql = "INSERT INTO evo(datetime, servers, users, mau) VALUES(%s,%s,%s,%s)"
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, (now, evo_servers, evo_users, evo_mau))
conn.commit()
cur.close()
except (Exception, psycopg2.DatabaseError) as error:
print(error)
finally:
if conn is not None:
conn.close()
##############################################################################
# get world's last update datetime
last_update = db.last_world_datetime()
conn = None
# get max servers and mau
try:
max_servers, max_mau = db.max()
conn = psycopg2.connect(database=fediverse_db, user=fediverse_db_user, password="", host="/var/run/postgresql", port="5432")
cur = conn.cursor()
cur.execute("select updated_at from world order by updated_at desc limit 1")
row = cur.fetchone()
last_update = row[0]
last_update = last_update.strftime('%m/%d/%Y, %H:%M:%S')
cur.close()
except (Exception, psycopg2.DatabaseError) as error:
print(error)
finally:
if conn is not None:
conn.close()
##############################################################################
# get max servers, users and mau
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("select MAX(total_servers) from totals")
row = cur.fetchone()
if row is not None:
max_servers = row[0]
else:
max_servers = 0
cur.execute("select MAX(total_users) from totals")
row = cur.fetchone()
if row is not None:
max_users = row[0]
else:
max_users = 0
cur.execute("select MAX(total_mau) from totals")
row = cur.fetchone()
if row is not None:
max_mau = row[0]
else:
max_mau = 0
cur.close()
except (Exception, psycopg2.DatabaseError) as error:
print(error)
finally:
if conn is not None:
conn.close()
###############################################################################
# get plots
servers_plots, mau_plots, global_week, global_servers, global_users, global_mau = db.get_plots()
servers_plots = []
users_plots = []
mau_plots = []
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("select total_servers, total_users, total_mau from totals order by datetime desc limit 14")
rows = cur.fetchall()
for row in rows:
servers_plots.append(row[0])
users_plots.append(row[1])
mau_plots.append(row[2])
cur.close()
except (Exception, psycopg2.DatabaseError) as error:
print(error)
finally:
if conn is not None:
conn.close()
###############################################################################
# generate graphs
@ -272,6 +845,22 @@ if __name__ == '__main__':
plt.close()
plt.plot([-6, -5, -4, -3, -2, -1, 0], [users_plots[6], users_plots[5], users_plots[4], users_plots[3], users_plots[2], users_plots[1], users_plots[0]], marker='o', color='royalblue')
plt.plot([-6, -5, -4, -3, -2, -1, 0], [max_users, max_users, max_users, max_users, max_users, max_users, max_users], color='red')
plt.title('fediverse: total registered users (max: ' + str(f"{max_users:,}" + ')'), loc='right', color='royalblue')
plt.legend(('users', 'max'), shadow=True, loc=(0.01, 0.80), handlelength=1.5, fontsize=10)
plt.xlabel('Last seven days')
plt.ylabel('Registered users')
plt.savefig('users.png')
plt.close()
plt.plot([-6, -5, -4, -3, -2, -1, 0], [mau_plots[6], mau_plots[5], mau_plots[4], mau_plots[3], mau_plots[2], mau_plots[1], mau_plots[0]], marker='o', color='royalblue')
plt.plot([-6, -5, -4, -3, -2, -1, 0], [max_mau, max_mau, max_mau, max_mau, max_mau, max_mau, max_mau], color='red')
@ -288,39 +877,8 @@ if __name__ == '__main__':
plt.close()
df = pd.DataFrame({'date': np.array(global_week),
#'servers': np.array(global_servers),
'users': np.array(global_users),
'mau': np.array(global_mau)})
df['date'] = pd.to_datetime(df['date'])
fig, ax = plt.subplots()
ax.plot(df.date, df.users, label='Registered', color='orange')
ax.plot(df.date, df.mau, label='MAU', color='blue')
plt.tick_params(rotation=45)
ax.set_title("fediverse's registered and Monthly Active Users")
ax.set_xlabel('weeks')
ax.set_ylabel('users')
ax.grid(visible=True)
ax.legend(title='Users')
ax.yaxis.set_major_formatter(y_formatter)
plt.savefig('global.png')
plt.close()
###############################################################################
# P O S T !
# T O O T !
toot_text = "#fediverse alive servers stats" + " \n"
@ -338,6 +896,18 @@ if __name__ == '__main__':
toot_text += "max: " + str(f"{max_servers:,}") + "\n"
#if evo_users >= 0:
# toot_text += "total users: " + str(f"{total_users:,}") + " (+"+ str(f"{evo_users:,}") + ") \n"
# toot_text += "max: " + str(f"{max_users:,}") + "\n"
#elif evo_users < 0:
# toot_text += "total users: " + str(f"{total_users:,}") + " ("+ str(f"{evo_users:,}") + ") \n"
# toot_text += "max: " + str(f"{max_users:,}") + "\n"
if evo_mau >= 0:
toot_text += "total MAU: " + str(f"{total_mau:,}") + " (+"+ str(f"{evo_mau:,}") + ") \n"
@ -350,7 +920,7 @@ if __name__ == '__main__':
toot_text += "max: " + str(f"{max_mau:,}") + "\n"
toot_text += "\ntop ten (MAU / servers):\n\n"
toot_text += "\ntop ten (soft MAU servers):\n\n"
i = 0
@ -358,17 +928,15 @@ if __name__ == '__main__':
project_soft = soft_total_project[i]
#project_users = soft_total_users[i]
project_mau = soft_total_mau[i]
project_servers = soft_total_servers[i]
len_pr_soft = len(project_soft)
if project_soft == 'ativity-relay':
project_soft = 'activityrelay'
toot_text += f":{project_soft}: {project_mau:,} / {project_servers:,}\n"
toot_text += f":{project_soft}: {project_mau:,} {project_servers:,}\n"
i += 1
@ -378,14 +946,18 @@ if __name__ == '__main__':
servers_image_id = mastodon.media_post('servers.png', "image/png", description='servers graph').id
#users_image_id = mastodon.media_post('users.png', "image/png", description='users graph').id
mau_image_id = mastodon.media_post('mau.png', "image/png", description='MAU graph').id
global_image_id = mastodon.media_post('global.png', "image/png", description='global graph').id
mastodon.status_post(toot_text, in_reply_to_id=None, media_ids={servers_image_id, mau_image_id})
mastodon.status_post(toot_text, in_reply_to_id=None, media_ids={servers_image_id, mau_image_id, global_image_id})
db.delete_dead_servers()
delete_dead_servers()
finish = datetime.now()
db.save_time(program, start, finish)
save_time(program, start, finish)
else:
usage()

Veure arxiu

@ -4,15 +4,13 @@ import os
import json
import sys
import os.path
from setup import Setup
from database import Database
import psycopg2
import requests
import urllib3
import socket
import ray
import pdb
#ray.init(num_cpus = 25) # Specify this system CPUs.
ray.init(num_cpus = 32) # Specify this system CPUs.
from ray.exceptions import (
RaySystemError,
@ -21,18 +19,18 @@ from ray.exceptions import (
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'
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?'
]
def is_json(myjson):
@ -43,6 +41,45 @@ def is_json(myjson):
return False
return True
def write_api(server, software, users, alive, api, soft_version):
fediverse_db, fediverse_db_user = get_db_config()
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:
conn = psycopg2.connect(database=fediverse_db, user=fediverse_db_user, password="", host="/var/run/postgresql", port="5432")
cur = conn.cursor()
print(f'Writing {server} nodeinfo data...')
cur.execute(insert_sql, (server, now, software, users, alive, api, soft_version))
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)
)
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()
@ray.remote
def getsoft(server):
@ -61,9 +98,11 @@ def getsoft(server):
url = 'https://' + server
user_agent = {'User-agent': "fediverse's stats (fediverse@mastodont.cat)"}
try:
response = requests.get(url + '/.well-known/nodeinfo', headers = setup.user_agent, timeout=3)
response = requests.get(url + '/.well-known/nodeinfo', headers = user_agent, timeout=3)
if response.status_code == 200:
@ -71,17 +110,11 @@ def getsoft(server):
response_json = response.json()
if len(response_json['links']) == 1:
nodeinfo = response_json['links'][0]['href'].replace(f'https://{server}','')
elif len(response_json['links']) == 2:
nodeinfo = response_json['links'][1]['href'].replace(f'https://{server}','')
try:
nodeinfo_data = requests.get(url + nodeinfo, headers = setup.user_agent, timeout=3)
nodeinfo_data = requests.get(nodeinfo, headers = user_agent, timeout=3)
if nodeinfo_data.status_code == 200:
@ -91,7 +124,7 @@ def getsoft(server):
else:
print(f"{nodeinfo} not responding: error code {nodeinfo_data.status_code}")
print(f"Server {server}'s nodeinfo not responding: error code {nodeinfo_data.status_code}")
except:
@ -99,7 +132,7 @@ def getsoft(server):
except:
print(f'{server} is not responding: error code {response.status_code}')
print(f'Server {server} not responding: error code {response.status_code}')
print('*********************************************************************')
pass
@ -109,9 +142,7 @@ def getsoft(server):
try:
response = requests.get(url + api, headers = setup.user_agent, timeout=3)
if response.status_code == 200:
response = requests.get(url + api, headers = user_agent, timeout=3)
if is_json(response.text):
@ -200,13 +231,12 @@ def getsoft(server):
soft = nodeinfo_json['software']['name']
soft = soft.lower()
soft_version = nodeinfo_json['software']['version']
users = nodeinfo_json.get('usage').get('users').get('total') or '0'
users = nodeinfo_json['usage']['users']['total']
if users > 1000000:
return
alive = True
db.write_api(server, soft, users, alive, nodeinfo, soft_version)
write_api(server, soft, users, alive, nodeinfo, soft_version)
print(f"Server {server} ({soft} {soft_version}) is alive!")
print('*********************************************************************')
@ -227,12 +257,11 @@ def getsoft(server):
users = nodeinfo_json['usage']['users']['total']
if users > 1000000:
return
alive = True
if soft == 'socialhome':
db.write_api(server, soft, users, alive, nodeinfo, soft_version)
write_api(server, soft, users, alive, nodeinfo, soft_version)
print('*********************************************************************')
print(f"Server {serve}r ({soft} {soft_version}) is alive!")
@ -270,7 +299,7 @@ def getsoft(server):
alive = True
db.write_api(server, soft, users, alive, nodeinfo, soft_version)
write_api(server, soft, users, alive, nodeinfo, soft_version)
print('*********************************************************************')
print(f"Server {server} ({soft}) is alive!")
@ -292,6 +321,106 @@ def getsoft(server):
print(f'Server {server} is dead')
print('*********************************************************************')
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
cur.execute("select server from world where checked='f'")
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()
def get_parameter(parameter, file_path):
# Check if secrets file exists
if not os.path.isfile(file_path):
print("File %s not found, exiting." % file_path)
sys.exit(0)
# Find parameter in file
with open(file_path) as f:
for line in f:
if line.startswith(parameter):
return line.replace(parameter + ":", "").strip()
# Cannot find parameter, exit
print(file_path + " Missing parameter %s " % parameter)
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
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)
###############################################################################
# main
@ -299,13 +428,7 @@ if __name__ == '__main__':
## name: fetchservers.py
setup = Setup()
db = Database()
res = requests.get('https://' + 'mastodon.social' + setup.peers_api, headers = setup.user_agent, timeout=3)
hostname_peers = res.json()
fediverse_db, fediverse_db_user = get_db_config()
start = datetime.now()
@ -313,19 +436,19 @@ if __name__ == '__main__':
finish = start
db.save_time(program, start, finish)
save_time(program, start, finish)
now = start
#world_servers = db.get_world_servers()
mastodon_hostname = get_config()
world_servers = get_world_servers()
ray_start = time.time()
try:
#results = ray.get([getsoft.remote(server) for server in world_servers])
results = ray.get([getsoft.remote(server) for server in hostname_peers])
#[getsoft(server) for server in world_servers]
results = ray.get([getsoft.remote(server) for server in world_servers])
print(f"duration = {time.time() - ray_start}.\nprocessed servers: {len(results)}")
@ -335,4 +458,4 @@ if __name__ == '__main__':
finish = datetime.now()
db.save_time(program, start, finish)
save_time(program, start, finish)

177
getpeers.py Normal file
Veure arxiu

@ -0,0 +1,177 @@
import os
import sys
import time
from datetime import datetime
import requests
import json
import psycopg2
import ray
import pdb
ray.init(num_cpus = 32) # Specify this system CPUs.
def write_server(server, federated_with):
insert_sql = "INSERT INTO world(server, federated_with, updated_at, saved_at, checked) VALUES(%s,%s,%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, (server, federated_with, now, now, 'f'))
print(f'writing {server} to world database')
conn.commit()
cur.close()
except (Exception, psycopg2.DatabaseError) as error:
print(error)
finally:
if conn is not None:
conn.close()
@ray.remote
def get_peers(peer):
try:
user_agent = {'User-agent': "fediverse's stats (fediverse@mastodont.cat)"}
domain_res = requests.get('https://' + peer + '/api/v1/instance', headers = user_agent, timeout=3)
domain_res_json = domain_res.json()
if domain_res.status_code == 200:
domain_uri = domain_res_json['uri'].replace('https://', '')
if domain_uri != peer:
print(f'{peer} is an aliased domain of {domain_uri}!')
else:
response = requests.get('https://' + peer + peers_api, headers = user_agent, timeout=3)
response_json = response.json()
if response.status_code == 200:
try:
print(f"Server: {peer}, federated with {str(len(response_json))} servers")
for peer_peer in response_json:
write_server(peer_peer, peer)
except:
pass
except:
pass
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()
def get_parameter( parameter, file_path ):
# Check if secrets file exists
if not os.path.isfile(file_path):
print("File %s not found, exiting."%file_path)
sys.exit(0)
# Find parameter in file
with open( file_path ) as f:
for line in f:
if line.startswith( parameter ):
return line.replace(parameter + ":", "").strip()
# Cannot find parameter, exit
print(file_path + " Missing parameter %s "%parameter)
sys.exit(0)
###############################################################################
# main
if __name__ == '__main__':
now = datetime.now()
peers_api = '/api/v1/instance/peers?'
# 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)
user_agent = {'User-agent': "fediverse's stats (fediverse@mastodont.cat)"}
res = requests.get('https://' + mastodon_hostname + peers_api, headers = user_agent, timeout=3)
hostname_peers = res.json()
start = datetime.now()
program = 'getpeers'
finish = start
save_time(program, start, finish)
for peer in hostname_peers:
write_server(peer, mastodon_hostname)
results = ray.get([get_peers.remote(server) for server in hostname_peers])
finish = datetime.now()
print(f"duration = {finish - start}.\nprocessed servers: {len(results)}")
save_time(program, start, finish)

Veure arxiu

@ -1,7 +1,9 @@
requests
psycopg2-binary
pytz
ray
Mastodon.py
matplotlib
pandas
Mastodon.py>=1.5.1
psycopg2-binary>=2.8.4
aiohttp>=3.6.2
aiodns>=2.0.0
matplotlib>=3.3.4
humanfriendly>=9.2
urllib3>=1.26.8
requests>=2.27.1
ray>=1.11.0

275
setup.py
Veure arxiu

@ -1,157 +1,226 @@
import os
import sys
from datetime import datetime
import pytz
import getpass
from mastodon import Mastodon
from mastodon.Mastodon import MastodonMalformedEventError, MastodonNetworkError, MastodonReadTimeout, MastodonAPIError, MastodonIllegalArgumentError
import pdb
import fileinput,re
import os
import sys
class Setup():
name = 'fediverse setup'
def __init__(self, config_file=None, mastodon_hostname=None, peers_api=None, user_agent=None, secrets_filepath=None, mastodon_app_token=None):
self.config_file = "config/config.txt"
self.mastodon_hostname = self.__get_parameter("mastodon_hostname", self.config_file)
self.peers_api = '/api/v1/instance/peers?'
self.user_agent = {'User-agent': "fediverse's stats (fediverse@mastodont.cat)"}
self.secrets_filepath = 'secrets/secrets.txt'
is_setup = self.__check_mastodon_setup(self)
if is_setup:
self.mastodon_app_token = self.__get_mastodon_parameter("mastodon_app_token", self.secrets_filepath)
else:
self.mastodon_app_token = self.mastodon_setup(self)
@staticmethod
def __check_mastodon_setup(self):
is_setup = False
if not os.path.isfile(self.secrets_filepath):
print(f"File {self.secrets_filepath} not found, running setup.")
else:
is_setup = True
return is_setup
@staticmethod
def mastodon_setup(self):
def create_dir():
if not os.path.exists('secrets'):
os.makedirs('secrets')
self.mastodon_user = input("Mastodon user login? ")
self.mastodon_password = input("Mastodon user password? ")
self.app_name = 'fediverse'
def create_file():
self.mastodon_app_token = self.mastodon_log_in()
if not os.path.exists('secrets/secrets.txt'):
with open('secrets/secrets.txt', 'w'): pass
print(secrets_filepath + " created!")
if not os.path.exists(self.secrets_filepath):
with open(self.secrets_filepath, 'w'): pass
print(f"{self.secrets_filepath} created!")
def create_config():
with open(self.secrets_filepath, 'a') as the_file:
print("Writing Mastodon parameters to " + self.secrets_filepath)
the_file.write(f'mastodon_app_token: {self.mastodon_app_token}')
if not os.path.exists('config'):
os.makedirs('config')
if not os.path.exists(config_filepath):
print(config_filepath + " created!")
with open('config/config.txt', 'w'): pass
return self.mastodon_app_token
def write_params():
def mastodon_log_in(self):
with open(secrets_filepath, 'a') as the_file:
print("Writing secrets parameter names to " + secrets_filepath)
the_file.write('uc_client_id: \n'+'uc_client_secret: \n'+'uc_access_token: \n')
token = ''
def write_config():
with open(config_filepath, 'a') as the_file:
the_file.write('mastodon_hostname: \n')
print("adding parameter name 'mastodon_hostname' to "+ config_filepath)
def read_client_lines(self):
client_path = 'app_clientcred.txt'
with open(client_path) as fp:
line = fp.readline()
cnt = 1
while line:
if cnt == 1:
print("Writing client id to " + secrets_filepath)
modify_file(secrets_filepath, "uc_client_id: ", value=line.rstrip())
elif cnt == 2:
print("Writing client secret to " + secrets_filepath)
modify_file(secrets_filepath, "uc_client_secret: ", value=line.rstrip())
line = fp.readline()
cnt += 1
def read_token_line(self):
token_path = 'app_usercred.txt'
with open(token_path) as fp:
line = fp.readline()
print("Writing access token to " + secrets_filepath)
modify_file(secrets_filepath, "uc_access_token: ", value=line.rstrip())
def read_config_line():
with open(config_filepath) as fp:
line = fp.readline()
modify_file(config_filepath, "mastodon_hostname: ", value=hostname)
modify_file(config_filepath, "bot_username: ", value=bot_username)
def log_in():
error = 0
try:
response = Mastodon.create_app(
self.app_name,
global hostname
hostname = input("Enter Mastodon hostname: ")
user_name = input("User name, ex. user@" + hostname +"? ")
user_password = getpass.getpass("User password? ")
bot_username = input("Bot's username, ex. fediverse: ")
app_name = input("This app name? ")
Mastodon.create_app(
app_name,
scopes=["read","write"],
to_file=None,
api_base_url=self.mastodon_hostname
to_file="app_clientcred.txt",
api_base_url=hostname
)
client_id = response[0]
client_secret = response[1]
mastodon = Mastodon(client_id = client_id, client_secret = client_secret, api_base_url = self.mastodon_hostname)
token = mastodon.log_in(
self.mastodon_user,
self.mastodon_password,
mastodon = Mastodon(client_id = "app_clientcred.txt", api_base_url = hostname)
mastodon.log_in(
user_name,
user_password,
scopes = ["read", "write"],
to_file = None
to_file = "app_usercred.txt"
)
print('Log in succesful!')
except MastodonIllegalArgumentError as i_error:
sys.stdout.write(f'\n{str(i_error)}\n')
error = 1
if os.path.exists("secrets/secrets.txt"):
print("Removing secrets/secrets.txt file..")
os.remove("secrets/secrets.txt")
if os.path.exists("app_clientcred.txt"):
print("Removing app_clientcred.txt file..")
os.remove("app_clientcred.txt")
sys.exit(i_error)
except MastodonNetworkError as n_error:
sys.stdout.write(f'\n{str(n_error)}\n')
error = 1
if os.path.exists("secrets/secrets.txt"):
print("Removing secrets/secrets.txt file..")
os.remove("secrets/secrets.txt")
if os.path.exists("app_clientcred.txt"):
print("Removing app_clientcred.txt file..")
os.remove("app_clientcred.txt")
sys.exit(n_error)
except MastodonReadTimeout as r_error:
sys.stdout.write(f'\n{str(r_error)}\n')
error = 1
if os.path.exists("secrets/secrets.txt"):
print("Removing secrets/secrets.txt file..")
os.remove("secrets/secrets.txt")
if os.path.exists("app_clientcred.txt"):
print("Removing app_clientcred.txt file..")
os.remove("app_clientcred.txt")
sys.exit(r_error)
except MastodonAPIError as a_error:
sys.stdout.write(f'\n{str(a_error)}\n')
error = 1
if os.path.exists("secrets/secrets.txt"):
print("Removing secrets/secrets.txt file..")
os.remove("secrets/secrets.txt")
if os.path.exists("app_clientcred.txt"):
print("Removing app_clientcred.txt file..")
os.remove("app_clientcred.txt")
sys.exit(a_error)
finally:
return token
if error == 0:
def __get_parameter(self, parameter, config_file):
create_dir()
create_file()
write_params()
client_path = 'app_clientcred.txt'
read_client_lines(client_path)
token_path = 'app_usercred.txt'
read_token_line(token_path)
if os.path.exists("app_clientcred.txt"):
print("Removing app_clientcred.txt temp file..")
os.remove("app_clientcred.txt")
if os.path.exists("app_usercred.txt"):
print("Removing app_usercred.txt temp file..")
os.remove("app_usercred.txt")
print("Secrets setup done!\n")
if not os.path.isfile(config_file):
print(f"File {config_file} not found..")
def modify_file(file_name,pattern,value=""):
self.mastodon_hostname = input("\nMastodon hostname: ")
fh=fileinput.input(file_name,inplace=True)
for line in fh:
replacement=pattern + value
line=re.sub(pattern,replacement,line)
sys.stdout.write(line)
fh.close()
self.__create_config(self)
self.__write_config(self)
def get_parameter( parameter, file_path ):
# Check if secrets file exists
if not os.path.isfile(file_path):
print("File %s not found, creating it."%file_path)
log_in()
with open( self.config_file ) as f:
# Find parameter in file
with open( file_path ) as f:
for line in f:
if line.startswith( parameter ):
return line.replace(parameter + ":", "").strip()
def __get_mastodon_parameter(self, parameter, secrets_filepath):
# Cannot find parameter, exit
print(file_path + " Missing parameter %s "%parameter)
sys.exit(0)
if not os.path.isfile(secrets_filepath):
print(f"File {secrets_filepath} not found..")
def get_hostname( parameter, config_filepath ):
# Check if secrets file exists
if not os.path.isfile(config_filepath):
print("File %s not found, creating it."%config_filepath)
create_config()
self.sign_in()
with open( self.secrets_filepath ) as f:
# Find parameter in file
with open( config_filepath ) as f:
for line in f:
if line.startswith( parameter ):
return line.replace(parameter + ":", "").strip()
@staticmethod
def __create_config(self):
# Cannot find parameter, exit
print(config_filepath + " Missing parameter %s "%parameter)
write_config()
read_config_line()
print("setup done!")
sys.exit(0)
if not os.path.exists('config'):
###############################################################################
# main
os.makedirs('config')
if __name__ == '__main__':
if not os.path.exists(self.config_file):
# Load secrets from secrets file
secrets_filepath = "secrets/secrets.txt"
uc_client_id = get_parameter("uc_client_id", secrets_filepath)
uc_client_secret = get_parameter("uc_client_secret", secrets_filepath)
uc_access_token = get_parameter("uc_access_token", secrets_filepath)
print(self.config_file + " created!")
with open(self.config_file, 'w'): pass
@staticmethod
def __write_config(self):
with open(self.config_file, 'a') as the_file:
the_file.write(f'mastodon_hostname: {self.mastodon_hostname}')
print(f"adding parameters to {self.config_file}\n")
# Load configuration from config file
config_filepath = "config/config.txt"
mastodon_hostname = get_hostname("mastodon_hostname", config_filepath)
bot_username = get_parameter("bot_username", config_filepath)