]> git.mxchange.org Git - fba.git/blob - fetch_blocks.py
One more case insensitivity fix
[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             if res.json()["software"]["name"] == "akkoma":
73                 return "pleroma"
74             else:
75                 return res.json()["software"]["name"]
76         elif res.status_code == 404:
77             res = get(f"https://{domain}/api/v1/instance", headers=headers, timeout=5)
78         if res.ok:
79             return "mastodon"
80     except:
81         return None
82
83
84 conn = sqlite3.connect("blocks.db")
85 c = conn.cursor()
86
87 c.execute(
88     "select domain, software from instances where software in ('pleroma', 'mastodon')"
89 )
90
91 for blocker, software in c.fetchall():
92     if software == "pleroma":
93         print(blocker)
94         try:
95             # Blocks
96             federation = get(
97                 f"https://{blocker}/nodeinfo/2.1.json", headers=headers, timeout=5
98             ).json()["metadata"]["federation"]
99             if "mrf_simple" in federation:
100                 for block_level, blocks in (
101                     {**federation["mrf_simple"],
102                     **{"quarantined_instances": federation["quarantined_instances"]}}
103                 ).items():
104                     for blocked in blocks:
105                         if blocked == "":
106                             continue
107                         blocked == blocked.lower()
108                         c.execute(
109                             "select domain from instances where domain = ?", (blocked,)
110                         )
111                         if c.fetchone() == None:
112                             c.execute(
113                                 "insert into instances select ?, ?, ?",
114                                 (blocked, get_hash(blocked), get_type(blocked)),
115                             )
116                         c.execute(
117                             "select * from blocks where blocker = ? and blocked = ? and block_level = ?",
118                             (blocker, blocked, block_level),
119                         )
120                         if c.fetchone() == None:
121                             c.execute(
122                                 "insert into blocks select ?, ?, '', ?",
123                                 (blocker, blocked, block_level),
124                             )
125             conn.commit()
126             # Reasons
127             if "mrf_simple_info" in federation:
128                 for block_level, info in (
129                     {**federation["mrf_simple_info"],
130                     **(federation["quarantined_instances_info"]
131                     if "quarantined_instances_info" in federation
132                     else {})}
133                 ).items():
134                     for blocked, reason in info.items():
135                         blocked == blocked.lower()
136                         c.execute(
137                             "update blocks set reason = ? where blocker = ? and blocked = ? and block_level = ?",
138                             (reason["reason"], blocker, blocked, block_level),
139                         )
140             conn.commit()
141         except Exception as e:
142             print("error:", e, blocker)
143     elif software == "mastodon":
144         print(blocker)
145         try:
146             json = get_mastodon_blocks(blocker)
147             for block_level, blocks in json.items():
148                 for instance in blocks:
149                     blocked, blocked_hash, reason = instance.values()
150                     blocked == blocked.lower()
151                     if blocked.count("*") <= 1:
152                         c.execute(
153                             "select hash from instances where hash = ?", (blocked_hash,)
154                         )
155                         if c.fetchone() == None:
156                             c.execute(
157                                 "insert into instances select ?, ?, ?",
158                                 (blocked, get_hash(blocked), get_type(blocked)),
159                             )
160                     c.execute(
161                         "select * from blocks where blocker = ? and blocked = ? and block_level = ?",
162                         (blocker, blocked, block_level),
163                     )
164                     if c.fetchone() == None:
165                         c.execute(
166                             "insert into blocks select ?, ?, ?, ?",
167                             (
168                                 blocker,
169                                 blocked if blocked.count("*") <= 1 else blocked_hash,
170                                 reason,
171                                 block_level,
172                             ),
173                         )
174             conn.commit()
175         except Exception as e:
176             print("error:", e, blocker)
177 conn.close()