fetch_blocks.py

   1 from requests import get
   2 from hashlib import sha256
   3 import sqlite3
   4 from bs4 import BeautifulSoup
   5
   6 headers = {
   7     "user-agent": "Mozilla/5.0 (Windows NT 10.0; rv:91.0) Gecko/20100101 Firefox/91.0"
   8 }
   9
  10
  11 def get_mastodon_blocks(domain: str) -> dict:
  12     blocks = {
  13         "Suspended servers": [],
  14         "Filtered media": [],
  15         "Limited servers": [],
  16         "Silenced servers": [],
  17     }
  18
  19     translations = {
  20         "Gesperrte Server": "Suspended servers",
  21         "Gefilterte Medien": "Filtered media",
  22         "Stummgeschaltete Server": "Silenced servers",
  23         "停止済みのサーバー": "Suspended servers",
  24         "メディアを拒否しているサーバー": "Filtered media",
  25         "サイレンス済みのサーバー": "Silenced servers",
  26         "Serveurs suspendus": "Suspended servers",
  27         "Médias filtrés": "Filtered media",
  28         "Serveurs limités": "Silenced servers",
  29     }
  30
  31     try:
  32         doc = BeautifulSoup(
  33             get(f"https://{domain}/about/more", headers=headers, timeout=5).text,
  34             "html.parser",
  35         )
  36     except:
  37         return {}
  38
  39     for header in doc.find_all("h3"):
  40         for line in header.find_next_siblings("table")[0].find_all("tr")[1:]:
  41             header_text = header.text
  42             if header_text in translations:
  43                     header_text = translations[header_text]
  44             if header_text in blocks:
  45                 blocks[header_text].append(
  46                     {
  47                         "domain": line.find("span").text,
  48                         "hash": line.find("span")["title"][9:],
  49                         "reason": line.find_all("td")[1].text.strip(),
  50                     }
  51                 )
  52     return {
  53         "reject": blocks["Suspended servers"],
  54         "media_removal": blocks["Filtered media"],
  55         "federated_timeline_removal": blocks["Limited servers"]
  56         + blocks["Silenced servers"],
  57     }
  58
  59
  60 def get_hash(domain: str) -> str:
  61     return sha256(domain.encode("utf-8")).hexdigest()
  62
  63
  64 def get_type(domain: str) -> str:
  65     try:
  66         res = get(f"https://{domain}/nodeinfo/2.1.json", headers=headers, timeout=5)
  67         if res.status_code == 404:
  68             res = get(f"https://{domain}/nodeinfo/2.0.json", headers=headers, timeout=5)
  69         if res.ok and "text/html" in res.headers["content-type"]:
  70             res = get(f"https://{domain}/nodeinfo/2.1", headers=headers, timeout=5)
  71         if res.ok:
  72             if res.json()["software"]["name"] == "akkoma":
  73                 return "pleroma"
  74             elif res.json()["software"]["name"] == "hometown":
  75                 return "mastodon"
  76             else:
  77                 return res.json()["software"]["name"]
  78         elif res.status_code == 404:
  79             res = get(f"https://{domain}/api/v1/instance", headers=headers, timeout=5)
  80         if res.ok:
  81             return "mastodon"
  82     except:
  83         return None
  84
  85
  86 conn = sqlite3.connect("blocks.db")
  87 c = conn.cursor()
  88
  89 c.execute(
  90     "select domain, software from instances where software in ('pleroma', 'mastodon')"
  91 )
  92
  93 for blocker, software in c.fetchall():
  94     if software == "pleroma":
  95         print(blocker)
  96         try:
  97             # Blocks
  98             federation = get(
  99                 f"https://{blocker}/nodeinfo/2.1.json", headers=headers, timeout=5
 100             ).json()["metadata"]["federation"]
 101             if "mrf_simple" in federation:
 102                 for block_level, blocks in (
 103                     {**federation["mrf_simple"],
 104                     **{"quarantined_instances": federation["quarantined_instances"]}}
 105                 ).items():
 106                     for blocked in blocks:
 107                         if blocked == "":
 108                             continue
 109                         blocked == blocked.lower()
 110                         blocker == blocker.lower()
 111                         c.execute(
 112                             "select domain from instances where domain = ?", (blocked,)
 113                         )
 114                         if c.fetchone() == None:
 115                             c.execute(
 116                                 "insert into instances select ?, ?, ?",
 117                                 (blocked, get_hash(blocked), get_type(blocked)),
 118                             )
 119                         c.execute(
 120                             "select * from blocks where blocker = ? and blocked = ? and block_level = ?",
 121                             (blocker, blocked, block_level),
 122                         )
 123                         if c.fetchone() == None:
 124                             c.execute(
 125                                 "insert into blocks select ?, ?, '', ?",
 126                                 (blocker, blocked, block_level),
 127                             )
 128             conn.commit()
 129             # Reasons
 130             if "mrf_simple_info" in federation:
 131                 for block_level, info in (
 132                     {**federation["mrf_simple_info"],
 133                     **(federation["quarantined_instances_info"]
 134                     if "quarantined_instances_info" in federation
 135                     else {})}
 136                 ).items():
 137                     for blocked, reason in info.items():
 138                         blocker == blocker.lower()
 139                         blocked == blocked.lower()
 140                         c.execute(
 141                             "update blocks set reason = ? where blocker = ? and blocked = ? and block_level = ?",
 142                             (reason["reason"], blocker, blocked, block_level),
 143                         )
 144             conn.commit()
 145         except Exception as e:
 146             print("error:", e, blocker)
 147     elif software == "mastodon":
 148         print(blocker)
 149         try:
 150             json = get_mastodon_blocks(blocker)
 151             for block_level, blocks in json.items():
 152                 for instance in blocks:
 153                     blocked, blocked_hash, reason = instance.values()
 154                     blocked == blocked.lower()
 155                     blocker == blocker.lower()
 156                     if blocked.count("*") <= 1:
 157                         c.execute(
 158                             "select hash from instances where hash = ?", (blocked_hash,)
 159                         )
 160                         if c.fetchone() == None:
 161                             c.execute(
 162                                 "insert into instances select ?, ?, ?",
 163                                 (blocked, get_hash(blocked), get_type(blocked)),
 164                             )
 165                     c.execute(
 166                         "select * from blocks where blocker = ? and blocked = ? and block_level = ?",
 167                         (blocker, blocked, block_level),
 168                     )
 169                     if c.fetchone() == None:
 170                         c.execute(
 171                             "insert into blocks select ?, ?, ?, ?",
 172                             (
 173                                 blocker,
 174                                 blocked if blocked.count("*") <= 1 else blocked_hash,
 175                                 reason,
 176                                 block_level,
 177                             ),
 178                         )
 179             conn.commit()
 180         except Exception as e:
 181             print("error:", e, blocker)
 182 conn.close()