]> git.mxchange.org Git - fba.git/blob - fetch_blocks.py
We do a little trolling
[fba.git] / 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             return res.json()["software"]["name"]
73         elif res.status_code == 404:
74             res = get(f"https://{domain}/api/v1/instance", headers=headers, timeout=5)
75         if res.ok:
76             return "mastodon"
77     except:
78         return None
79
80
81 conn = sqlite3.connect("blocks.db")
82 c = conn.cursor()
83
84 c.execute(
85     "select domain, software from instances where software in ('pleroma', 'mastodon')"
86 )
87
88 for blocker, software in c.fetchall():
89     if software == "pleroma":
90         print(blocker)
91         try:
92             # Blocks
93             federation = get(
94                 f"https://{blocker}/nodeinfo/2.1.json", headers=headers, timeout=5
95             ).json()["metadata"]["federation"]
96             if "mrf_simple" in federation:
97                 for block_level, blocks in (
98                     {**federation["mrf_simple"],
99                     **{"quarantined_instances": federation["quarantined_instances"]}}
100                 ).items():
101                     for blocked in blocks:
102                         if blocked == "":
103                             continue
104                         c.execute(
105                             "select domain from instances where domain = ?", (blocked,)
106                         )
107                         if c.fetchone() == None:
108                             c.execute(
109                                 "insert into instances select ?, ?, ?",
110                                 (blocked, get_hash(blocked), get_type(blocked)),
111                             )
112                         c.execute(
113                             "select * from blocks where blocker = ? and blocked = ? and block_level = ?",
114                             (blocker, blocked, block_level),
115                         )
116                         if c.fetchone() == None:
117                             c.execute(
118                                 "insert into blocks select ?, ?, '', ?",
119                                 (blocker, blocked, block_level),
120                             )
121             conn.commit()
122             # Reasons
123             if "mrf_simple_info" in federation:
124                 for block_level, info in (
125                     {**federation["mrf_simple_info"],
126                     **(federation["quarantined_instances_info"]
127                     if "quarantined_instances_info" in federation
128                     else {})}
129                 ).items():
130                     for blocked, reason in info.items():
131                         c.execute(
132                             "update blocks set reason = ? where blocker = ? and blocked = ? and block_level = ?",
133                             (reason["reason"], blocker, blocked, block_level),
134                         )
135             conn.commit()
136         except Exception as e:
137             print("error:", e, blocker)
138     elif software == "mastodon":
139         print(blocker)
140         try:
141             json = get_mastodon_blocks(blocker)
142             for block_level, blocks in json.items():
143                 for instance in blocks:
144                     blocked, blocked_hash, reason = instance.values()
145                     if blocked.count("*") <= 1:
146                         c.execute(
147                             "select hash from instances where hash = ?", (blocked_hash,)
148                         )
149                         if c.fetchone() == None:
150                             c.execute(
151                                 "insert into instances select ?, ?, ?",
152                                 (blocked, get_hash(blocked), get_type(blocked)),
153                             )
154                     c.execute(
155                         "select * from blocks where blocker = ? and blocked = ? and block_level = ?",
156                         (blocker, blocked, block_level),
157                     )
158                     if c.fetchone() == None:
159                         c.execute(
160                             "insert into blocks select ?, ?, ?, ?",
161                             (
162                                 blocker,
163                                 blocked if blocked.count("*") <= 1 else blocked_hash,
164                                 reason,
165                                 block_level,
166                             ),
167                         )
168             conn.commit()
169         except Exception as e:
170             print("error:", e, blocker)
171 conn.close()