]> git.mxchange.org Git - fba.git/blob - fetch_blocks.py
Add aonther nodeinfo endpoint used by Friendica
[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 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", headers=headers, timeout=5)
92         if res.status_code == 404:
93             res = get(f"https://{domain}/nodeinfo/2.0.json", headers=headers, timeout=5)
94         if res.ok and "text/html" in res.headers["content-type"]:
95             res = get(f"https://{domain}/nodeinfo/2.1", headers=headers, timeout=5)
96         if res.ok:
97             if res.json()["software"]["name"] == "akkoma":
98                 return "pleroma"
99             elif res.json()["software"]["name"] == "hometown":
100                 return "mastodon"
101             else:
102                 return res.json()["software"]["name"]
103         elif res.status_code == 404:
104             res = get(f"https://{domain}/api/v1/instance", headers=headers, timeout=5)
105         if res.ok:
106             return "mastodon"
107     except:
108         return None
109
110
111 conn = sqlite3.connect("blocks.db")
112 c = conn.cursor()
113
114 c.execute(
115     "select domain, software from instances where software in ('pleroma', 'mastodon')"
116 )
117
118 for blocker, software in c.fetchall():
119     if software == "pleroma":
120         print(blocker)
121         try:
122             # Blocks
123             federation = get(
124                 f"https://{blocker}/nodeinfo/2.1.json", headers=headers, timeout=5
125             ).json()["metadata"]["federation"]
126             if "mrf_simple" in federation:
127                 for block_level, blocks in (
128                     {**federation["mrf_simple"],
129                     **{"quarantined_instances": federation["quarantined_instances"]}}
130                 ).items():
131                     for blocked in blocks:
132                         if blocked == "":
133                             continue
134                         blocked == blocked.lower()
135                         blocker == blocker.lower()
136                         c.execute(
137                             "select domain from instances where domain = ?", (blocked,)
138                         )
139                         if c.fetchone() == None:
140                             c.execute(
141                                 "insert into instances select ?, ?, ?",
142                                 (blocked, get_hash(blocked), get_type(blocked)),
143                             )
144                         c.execute(
145                             "select * from blocks where blocker = ? and blocked = ? and block_level = ?",
146                             (blocker, blocked, block_level),
147                         )
148                         if c.fetchone() == None:
149                             c.execute(
150                                 "insert into blocks select ?, ?, '', ?",
151                                 (blocker, blocked, block_level),
152                             )
153             conn.commit()
154             # Reasons
155             if "mrf_simple_info" in federation:
156                 for block_level, info in (
157                     {**federation["mrf_simple_info"],
158                     **(federation["quarantined_instances_info"]
159                     if "quarantined_instances_info" in federation
160                     else {})}
161                 ).items():
162                     for blocked, reason in info.items():
163                         blocker == blocker.lower()
164                         blocked == blocked.lower()
165                         c.execute(
166                             "update blocks set reason = ? where blocker = ? and blocked = ? and block_level = ?",
167                             (reason["reason"], blocker, blocked, block_level),
168                         )
169             conn.commit()
170         except Exception as e:
171             print("error:", e, blocker)
172     elif software == "mastodon":
173         print(blocker)
174         try:
175             json = get_mastodon_blocks(blocker)
176             for block_level, blocks in json.items():
177                 for instance in blocks:
178                     blocked, blocked_hash, reason = instance.values()
179                     blocked == blocked.lower()
180                     blocker == blocker.lower()
181                     if blocked.count("*") <= 1:
182                         c.execute(
183                             "select hash from instances where hash = ?", (blocked_hash,)
184                         )
185                         if c.fetchone() == None:
186                             c.execute(
187                                 "insert into instances select ?, ?, ?",
188                                 (blocked, get_hash(blocked), get_type(blocked)),
189                             )
190                     c.execute(
191                         "select * from blocks where blocker = ? and blocked = ? and block_level = ?",
192                         (blocker, blocked if blocked.count("*") <= 1 else blocked_hash, block_level),
193                     )
194                     if c.fetchone() == None:
195                         c.execute(
196                             "insert into blocks select ?, ?, ?, ?",
197                             (
198                                 blocker,
199                                 blocked if blocked.count("*") <= 1 else blocked_hash,
200                                 reason,
201                                 block_level,
202                             ),
203                         )
204             conn.commit()
205         except Exception as e:
206             print("error:", e, blocker)
207     elif software == "friendica"
208         print(blocker)
209         try:
210             json = get_friendica_blocks(blocker)
211             for blocks in json.items():
212                 for instance in blocks:
213                     blocked, reason = instance.values()
214                     blocked == blocked.lower()
215                     blocker == blocker.lower()
216                     c.execute(
217                         "select domain from instances where domain = ?", (blocked,)
218                     )
219                     if c.fetchone() == None:
220                         c.execute(
221                             "insert into instances select ?, ?, ?",
222                             (blocked, get_hash(blocked), get_type(blocked)),
223                         )
224                     c.execute(
225                         "select * from blocks where blocker = ? and blocked = ?",
226                         (blocker, blocked),
227                     )
228                     if c.fetchone() == None:
229                         c.execute(
230                             "insert into blocks select ?, ?, ?, ?",
231                             (
232                                 blocker,
233                                 blocked,
234                                 reason,
235                                 "reject",
236                             ),
237                         )
238             conn.commit()
239         except Exception as e:
240             print("error:", e, blocker)
241 conn.close()