contrib/seeds/makeseeds.py

   1 #!/usr/bin/env python3
   2 # Copyright (c) 2013-2017 The Bitcoin Core developers
   3 # Distributed under the MIT software license, see the accompanying
   4 # file COPYING or http://www.opensource.org/licenses/mit-license.php.
   5 #
   6 # Generate seeds.txt from Pieter's DNS seeder
   7 #
   8
   9 NSEEDS=512
  10
  11 MAX_SEEDS_PER_ASN=2
  12
  13 MIN_BLOCKS = 337600
  14
  15 # These are hosts that have been observed to be behaving strangely (e.g.
  16 # aggressively connecting to every node).
  17 SUSPICIOUS_HOSTS = {
  18     "130.211.129.106", "178.63.107.226",
  19     "83.81.130.26", "88.198.17.7", "148.251.238.178", "176.9.46.6",
  20     "54.173.72.127", "54.174.10.182", "54.183.64.54", "54.194.231.211",
  21     "54.66.214.167", "54.66.220.137", "54.67.33.14", "54.77.251.214",
  22     "54.94.195.96", "54.94.200.247"
  23 }
  24
  25 import re
  26 import sys
  27 import dns.resolver
  28 import collections
  29
  30 PATTERN_IPV4 = re.compile(r"^((\d{1,3})\.(\d{1,3})\.(\d{1,3})\.(\d{1,3})):(\d+)$")
  31 PATTERN_IPV6 = re.compile(r"^\[([0-9a-z:]+)\]:(\d+)$")
  32 PATTERN_ONION = re.compile(r"^([abcdefghijklmnopqrstuvwxyz234567]{16}\.onion):(\d+)$")
  33 PATTERN_AGENT = re.compile(r"^(/Satoshi:0.12.(0|1|99)/|/Satoshi:0.13.(0|1|2|99)/)$")
  34
  35 def parseline(line):
  36     sline = line.split()
  37     if len(sline) < 11:
  38        return None
  39     m = PATTERN_IPV4.match(sline[0])
  40     sortkey = None
  41     ip = None
  42     if m is None:
  43         m = PATTERN_IPV6.match(sline[0])
  44         if m is None:
  45             m = PATTERN_ONION.match(sline[0])
  46             if m is None:
  47                 return None
  48             else:
  49                 net = 'onion'
  50                 ipstr = sortkey = m.group(1)
  51                 port = int(m.group(2))
  52         else:
  53             net = 'ipv6'
  54             if m.group(1) in ['::']: # Not interested in localhost
  55                 return None
  56             ipstr = m.group(1)
  57             sortkey = ipstr # XXX parse IPv6 into number, could use name_to_ipv6 from generate-seeds
  58             port = int(m.group(2))
  59     else:
  60         # Do IPv4 sanity check
  61         ip = 0
  62         for i in range(0,4):
  63             if int(m.group(i+2)) < 0 or int(m.group(i+2)) > 255:
  64                 return None
  65             ip = ip + (int(m.group(i+2)) << (8*(3-i)))
  66         if ip == 0:
  67             return None
  68         net = 'ipv4'
  69         sortkey = ip
  70         ipstr = m.group(1)
  71         port = int(m.group(6))
  72     # Skip bad results.
  73     if sline[1] == 0:
  74         return None
  75     # Extract uptime %.
  76     uptime30 = float(sline[7][:-1])
  77     # Extract Unix timestamp of last success.
  78     lastsuccess = int(sline[2])
  79     # Extract protocol version.
  80     version = int(sline[10])
  81     # Extract user agent.
  82     agent = sline[11][1:-1]
  83     # Extract service flags.
  84     service = int(sline[9], 16)
  85     # Extract blocks.
  86     blocks = int(sline[8])
  87     # Construct result.
  88     return {
  89         'net': net,
  90         'ip': ipstr,
  91         'port': port,
  92         'ipnum': ip,
  93         'uptime': uptime30,
  94         'lastsuccess': lastsuccess,
  95         'version': version,
  96         'agent': agent,
  97         'service': service,
  98         'blocks': blocks,
  99         'sortkey': sortkey,
 100     }
 101
 102 def filtermultiport(ips):
 103     '''Filter out hosts with more nodes per IP'''
 104     hist = collections.defaultdict(list)
 105     for ip in ips:
 106         hist[ip['sortkey']].append(ip)
 107     return [value[0] for (key,value) in list(hist.items()) if len(value)==1]
 108
 109 # Based on Greg Maxwell's seed_filter.py
 110 def filterbyasn(ips, max_per_asn, max_total):
 111     # Sift out ips by type
 112     ips_ipv4 = [ip for ip in ips if ip['net'] == 'ipv4']
 113     ips_ipv6 = [ip for ip in ips if ip['net'] == 'ipv6']
 114     ips_onion = [ip for ip in ips if ip['net'] == 'onion']
 115
 116     # Filter IPv4 by ASN
 117     result = []
 118     asn_count = {}
 119     for ip in ips_ipv4:
 120         if len(result) == max_total:
 121             break
 122         try:
 123             asn = int([x.to_text() for x in dns.resolver.query('.'.join(reversed(ip['ip'].split('.'))) + '.origin.asn.cymru.com', 'TXT').response.answer][0].split('\"')[1].split(' ')[0])
 124             if asn not in asn_count:
 125                 asn_count[asn] = 0
 126             if asn_count[asn] == max_per_asn:
 127                 continue
 128             asn_count[asn] += 1
 129             result.append(ip)
 130         except:
 131             sys.stderr.write('ERR: Could not resolve ASN for "' + ip['ip'] + '"\n')
 132
 133     # TODO: filter IPv6 by ASN
 134
 135     # Add back non-IPv4
 136     result.extend(ips_ipv6)
 137     result.extend(ips_onion)
 138     return result
 139
 140 def main():
 141     lines = sys.stdin.readlines()
 142     ips = [parseline(line) for line in lines]
 143
 144     # Skip entries with valid address.
 145     ips = [ip for ip in ips if ip is not None]
 146     # Skip entries from suspicious hosts.
 147     ips = [ip for ip in ips if ip['ip'] not in SUSPICIOUS_HOSTS]
 148     # Enforce minimal number of blocks.
 149     ips = [ip for ip in ips if ip['blocks'] >= MIN_BLOCKS]
 150     # Require service bit 1.
 151     ips = [ip for ip in ips if (ip['service'] & 1) == 1]
 152     # Require at least 50% 30-day uptime.
 153     ips = [ip for ip in ips if ip['uptime'] > 50]
 154     # Require a known and recent user agent.
 155     ips = [ip for ip in ips if PATTERN_AGENT.match(ip['agent'])]
 156     # Sort by availability (and use last success as tie breaker)
 157     ips.sort(key=lambda x: (x['uptime'], x['lastsuccess'], x['ip']), reverse=True)
 158     # Filter out hosts with multiple bitcoin ports, these are likely abusive
 159     ips = filtermultiport(ips)
 160     # Look up ASNs and limit results, both per ASN and globally.
 161     ips = filterbyasn(ips, MAX_SEEDS_PER_ASN, NSEEDS)
 162     # Sort the results by IP address (for deterministic output).
 163     ips.sort(key=lambda x: (x['net'], x['sortkey']))
 164
 165     for ip in ips:
 166         if ip['net'] == 'ipv6':
 167             print('[%s]:%i' % (ip['ip'], ip['port']))
 168         else:
 169             print('%s:%i' % (ip['ip'], ip['port']))
 170
 171 if __name__ == '__main__':
 172     main()