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 def get_friendica_blocks(domain: str) -> dict:
  60     blocks = []
  61
  62     try:
  63         doc = BeautifulSoup(
  64             get(f"https://{domain}/friendica", headers=headers, timeout=5).text,
  65             "html.parser",
  66         )
  67     except:
  68         return {}
  69
  70     blocklist = doc.find(id="about_blocklist")
  71     for line in blocklist.find("table").find_all("tr")[1:]:
  72             blocks.append(
  73                 {
  74                     "domain": line.find_all("td")[0].text.strip(),
  75                     "reason": line.find_all("td")[1].text.strip()
  76                 }
  77             )
  78
  79     return {
  80         "reject": blocks
  81     }
  82
  83 def get_hash(domain: str) -> str:
  84     return sha256(domain.encode("utf-8")).hexdigest()
  85
  86
  87 def get_type(domain: str) -> str:
  88     try:
  89         res = get(f"https://{domain}/nodeinfo/2.1.json", headers=headers, timeout=5)
  90         if res.status_code == 404:
  91             res = get(f"https://{domain}/nodeinfo/2.0.json", headers=headers, timeout=5)
  92         if res.ok and "text/html" in res.headers["content-type"]:
  93             res = get(f"https://{domain}/nodeinfo/2.1", headers=headers, timeout=5)
  94         if res.ok:
  95             if res.json()["software"]["name"] == "akkoma":
  96                 return "pleroma"
  97             elif res.json()["software"]["name"] == "hometown":
  98                 return "mastodon"
  99             else:
 100                 return res.json()["software"]["name"]
 101         elif res.status_code == 404:
 102             res = get(f"https://{domain}/api/v1/instance", headers=headers, timeout=5)
 103         if res.ok:
 104             return "mastodon"
 105     except:
 106         return None
 107
 108
 109 conn = sqlite3.connect("blocks.db")
 110 c = conn.cursor()
 111
 112 c.execute(
 113     "select domain, software from instances where software in ('pleroma', 'mastodon')"
 114 )
 115
 116 for blocker, software in c.fetchall():
 117     if software == "pleroma":
 118         print(blocker)
 119         try:
 120             # Blocks
 121             federation = get(
 122                 f"https://{blocker}/nodeinfo/2.1.json", headers=headers, timeout=5
 123             ).json()["metadata"]["federation"]
 124             if "mrf_simple" in federation:
 125                 for block_level, blocks in (
 126                     {**federation["mrf_simple"],
 127                     **{"quarantined_instances": federation["quarantined_instances"]}}
 128                 ).items():
 129                     for blocked in blocks:
 130                         if blocked == "":
 131                             continue
 132                         blocked == blocked.lower()
 133                         blocker == blocker.lower()
 134                         c.execute(
 135                             "select domain from instances where domain = ?", (blocked,)
 136                         )
 137                         if c.fetchone() == None:
 138                             c.execute(
 139                                 "insert into instances select ?, ?, ?",
 140                                 (blocked, get_hash(blocked), get_type(blocked)),
 141                             )
 142                         c.execute(
 143                             "select * from blocks where blocker = ? and blocked = ? and block_level = ?",
 144                             (blocker, blocked, block_level),
 145                         )
 146                         if c.fetchone() == None:
 147                             c.execute(
 148                                 "insert into blocks select ?, ?, '', ?",
 149                                 (blocker, blocked, block_level),
 150                             )
 151             conn.commit()
 152             # Reasons
 153             if "mrf_simple_info" in federation:
 154                 for block_level, info in (
 155                     {**federation["mrf_simple_info"],
 156                     **(federation["quarantined_instances_info"]
 157                     if "quarantined_instances_info" in federation
 158                     else {})}
 159                 ).items():
 160                     for blocked, reason in info.items():
 161                         blocker == blocker.lower()
 162                         blocked == blocked.lower()
 163                         c.execute(
 164                             "update blocks set reason = ? where blocker = ? and blocked = ? and block_level = ?",
 165                             (reason["reason"], blocker, blocked, block_level),
 166                         )
 167             conn.commit()
 168         except Exception as e:
 169             print("error:", e, blocker)
 170     elif software == "mastodon":
 171         print(blocker)
 172         try:
 173             json = get_mastodon_blocks(blocker)
 174             for block_level, blocks in json.items():
 175                 for instance in blocks:
 176                     blocked, blocked_hash, reason = instance.values()
 177                     blocked == blocked.lower()
 178                     blocker == blocker.lower()
 179                     if blocked.count("*") <= 1:
 180                         c.execute(
 181                             "select hash from instances where hash = ?", (blocked_hash,)
 182                         )
 183                         if c.fetchone() == None:
 184                             c.execute(
 185                                 "insert into instances select ?, ?, ?",
 186                                 (blocked, get_hash(blocked), get_type(blocked)),
 187                             )
 188                     c.execute(
 189                         "select * from blocks where blocker = ? and blocked = ? and block_level = ?",
 190                         (blocker, blocked if blocked.count("*") <= 1 else blocked_hash, block_level),
 191                     )
 192                     if c.fetchone() == None:
 193                         c.execute(
 194                             "insert into blocks select ?, ?, ?, ?",
 195                             (
 196                                 blocker,
 197                                 blocked if blocked.count("*") <= 1 else blocked_hash,
 198                                 reason,
 199                                 block_level,
 200                             ),
 201                         )
 202             conn.commit()
 203         except Exception as e:
 204             print("error:", e, blocker)
 205     elif software == "friendica"
 206         print(blocker)
 207         try:
 208             json = get_friendica_blocks(blocker)
 209             for blocks in json.items():
 210                 for instance in blocks:
 211                     blocked, reason = instance.values()
 212                     blocked == blocked.lower()
 213                     blocker == blocker.lower()
 214                     c.execute(
 215                         "select domain from instances where domain = ?", (blocked,)
 216                     )
 217                     if c.fetchone() == None:
 218                         c.execute(
 219                             "insert into instances select ?, ?, ?",
 220                             (blocked, get_hash(blocked), get_type(blocked)),
 221                         )
 222                     c.execute(
 223                         "select * from blocks where blocker = ? and blocked = ?",
 224                         (blocker, blocked),
 225                     )
 226                     if c.fetchone() == None:
 227                         c.execute(
 228                             "insert into blocks select ?, ?, ?, ?",
 229                             (
 230                                 blocker,
 231                                 blocked,
 232                                 reason,
 233                                 "reject",
 234                             ),
 235                         )
 236             conn.commit()
 237         except Exception as e:
 238             print("error:", e, blocker)
 239 conn.close()