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": "fedi-block-api (https://gitlab.com/EnjuAihara/fedi-block-api)"
   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     try:
  20         doc = BeautifulSoup(
  21             get(f"https://{domain}/about/more", headers=headers, timeout=5).text,
  22             "html.parser",
  23         )
  24     except:
  25         return {}
  26
  27     for header in doc.find_all("h3"):
  28         for line in header.find_next_siblings("table")[0].find_all("tr")[1:]:
  29             if header.text in blocks:
  30                 blocks[header.text].append(
  31                     {
  32                         "domain": line.find("span").text,
  33                         "hash": line.find("span")["title"][9:],
  34                         "reason": line.find_all("td")[1].text.strip(),
  35                     }
  36                 )
  37     return {
  38         "reject": blocks["Suspended servers"],
  39         "media_removal": blocks["Filtered media"],
  40         "federated_timeline_removal": blocks["Limited servers"]
  41         + blocks["Silenced servers"],
  42     }
  43
  44
  45 def get_hash(domain: str) -> str:
  46     return sha256(domain.encode("utf-8")).hexdigest()
  47
  48
  49 def get_type(domain: str) -> str:
  50     try:
  51         res = get(f"https://{domain}/nodeinfo/2.1.json", headers=headers, timeout=5)
  52         if res.status_code == 404:
  53             res = get(f"https://{domain}/nodeinfo/2.0.json", headers=headers, timeout=5)
  54         if res.ok:
  55             return res.json()["software"]["name"]
  56         elif res.status_code == 404:
  57             res = get(f"https://{domain}/api/v1/instance", headers=headers, timeout=5)
  58         if res.ok:
  59             return "mastodon"
  60     except:
  61         return None
  62
  63
  64 conn = sqlite3.connect("blocks.db")
  65 c = conn.cursor()
  66
  67 c.execute(
  68     "select domain, software from instances where software in ('pleroma', 'mastodon')"
  69 )
  70
  71 for blocker, software in c.fetchall():
  72     if software == "pleroma":
  73         print(blocker)
  74         try:
  75             # Blocks
  76             c.execute("delete from blocks where blocker = ?", (blocker,))
  77             federation = get(
  78                 f"https://{blocker}/nodeinfo/2.1.json", headers=headers, timeout=5
  79             ).json()["metadata"]["federation"]
  80             if "mrf_simple" in federation:
  81                 for block_level, blocks in (
  82                     federation["mrf_simple"]
  83                     | {"quarantined_instances": federation["quarantined_instances"]}
  84                 ).items():
  85                     for blocked in blocks:
  86                         if blocked == "":
  87                             continue
  88                         c.execute(
  89                             "select domain from instances where domain = ?", (blocked,)
  90                         )
  91                         if c.fetchone() == None:
  92                             c.execute(
  93                                 "insert into instances select ?, ?, ?",
  94                                 (blocked, get_hash(blocked), get_type(blocked)),
  95                             )
  96                         c.execute(
  97                             "insert into blocks select ?, ?, '', ?",
  98                             (blocker, blocked, block_level),
  99                         )
 100             conn.commit()
 101             # Reasons
 102             if "mrf_simple_info" in federation:
 103                 for block_level, info in (
 104                     federation["mrf_simple_info"]
 105                     | federation["quarantined_instances_info"]
 106                     if "quarantined_instances_info" in federation
 107                     else {}
 108                 ).items():
 109                     for blocked, reason in info.items():
 110                         c.execute(
 111                             "update blocks set reason = ? where blocker = ? and blocked = ? and block_level = ?",
 112                             (reason["reason"], blocker, blocked, block_level),
 113                         )
 114             conn.commit()
 115         except Exception as e:
 116             print("error:", e, blocker)
 117     elif software == "mastodon":
 118         print(blocker)
 119         try:
 120             c.execute("delete from blocks where blocker = ?", (blocker,))
 121             json = get_mastodon_blocks(blocker)
 122             for block_level, blocks in json.items():
 123                 for instance in blocks:
 124                     blocked, blocked_hash, reason = instance.values()
 125                     if blocked.count("*") <= 1:
 126                         c.execute(
 127                             "select hash from instances where hash = ?", (blocked_hash,)
 128                         )
 129                         if c.fetchone() == None:
 130                             c.execute(
 131                                 "insert into instances select ?, ?, ?",
 132                                 (blocked, get_hash(blocked), get_type(blocked)),
 133                             )
 134                     c.execute(
 135                         "insert into blocks select ?, ?, ?, ?",
 136                         (
 137                             blocker,
 138                             blocked if blocked.count("*") <= 1 else blocked_hash,
 139                             reason,
 140                             block_level,
 141                         ),
 142                     )
 143             conn.commit()
 144         except Exception as e:
 145             print("error:", e, blocker)
 146 conn.close()