]> git.mxchange.org Git - fba.git/blob - fetch_blocks.py
e2b5a3838f680b6220808190adc846b22197ba1f
[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:102.0) Gecko/20100101 Firefox/102.0"
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         "Silenced instances": "Silenced servers",
21         "Suspended instances": "Suspended servers",
22         "Gesperrte Server": "Suspended servers",
23         "Gefilterte Medien": "Filtered media",
24         "Stummgeschaltete Server": "Silenced servers",
25         "停止済みのサーバー": "Suspended servers",
26         "メディアを拒否しているサーバー": "Filtered media",
27         "サイレンス済みのサーバー": "Silenced servers",
28         "Serveurs suspendus": "Suspended servers",
29         "Médias filtrés": "Filtered media",
30         "Serveurs limités": "Silenced servers",
31     }
32
33     try:
34         doc = BeautifulSoup(
35             get(f"https://{domain}/about/more", headers=headers, timeout=5).text,
36             "html.parser",
37         )
38     except:
39         return {}
40
41     for header in doc.find_all("h3"):
42         for line in header.find_next_siblings("table")[0].find_all("tr")[1:]:
43             header_text = header.text
44             if header_text in translations:
45                     header_text = translations[header_text]
46             if header_text in blocks:
47                 blocks[header_text].append(
48                     {
49                         "domain": line.find("span").text,
50                         "hash": line.find("span")["title"][9:],
51                         "reason": line.find_all("td")[1].text.strip(),
52                     }
53                 )
54     return {
55         "reject": blocks["Suspended servers"],
56         "media_removal": blocks["Filtered media"],
57         "federated_timeline_removal": blocks["Limited servers"]
58         + blocks["Silenced servers"],
59     }
60
61 def get_friendica_blocks(domain: str) -> dict:
62     blocks = []
63
64     try:
65         doc = BeautifulSoup(
66             get(f"https://{domain}/friendica", headers=headers, timeout=5).text,
67             "html.parser",
68         )
69     except:
70         return {}
71
72     blocklist = doc.find(id="about_blocklist")
73     for line in blocklist.find("table").find_all("tr")[1:]:
74             blocks.append(
75                 {
76                     "domain": line.find_all("td")[0].text.strip(),
77                     "reason": line.find_all("td")[1].text.strip()
78                 }
79             )
80
81     return {
82         "reject": blocks
83     }
84
85 def get_hash(domain: str) -> str:
86     return sha256(domain.encode("utf-8")).hexdigest()
87
88
89 def get_type(domain: str) -> str:
90     try:
91         res = get(f"https://{domain}/nodeinfo/2.1.json", headers=headers, timeout=5)
92         if res.status_code == 404:
93             res = get(f"https://{domain}/nodeinfo/2.0", headers=headers, timeout=5)
94         if res.status_code == 404:
95             res = get(f"https://{domain}/nodeinfo/2.0.json", headers=headers, timeout=5)
96         if res.ok and "text/html" in res.headers["content-type"]:
97             res = get(f"https://{domain}/nodeinfo/2.1", headers=headers, timeout=5)
98         if res.ok:
99             if res.json()["software"]["name"] == "akkoma":
100                 return "pleroma"
101             elif res.json()["software"]["name"] == "rebased":
102                 return "pleroma"
103             elif res.json()["software"]["name"] == "hometown":
104                 return "mastodon"
105             elif res.json()["software"]["name"] == "ecko":
106                 return "mastodon"
107             else:
108                 return res.json()["software"]["name"]
109         elif res.status_code == 404:
110             res = get(f"https://{domain}/api/v1/instance", headers=headers, timeout=5)
111         if res.ok:
112             return "mastodon"
113     except:
114         return None
115
116
117 conn = sqlite3.connect("blocks.db")
118 c = conn.cursor()
119
120 c.execute(
121     "select domain, software from instances where software in ('pleroma', 'mastodon', 'friendica')"
122 )
123
124 for blocker, software in c.fetchall():
125     if software == "pleroma":
126         print(blocker)
127         try:
128             # Blocks
129             federation = get(
130                 f"https://{blocker}/nodeinfo/2.1.json", headers=headers, timeout=5
131             ).json()["metadata"]["federation"]
132             if "mrf_simple" in federation:
133                 for block_level, blocks in (
134                     {**federation["mrf_simple"],
135                     **{"quarantined_instances": federation["quarantined_instances"]}}
136                 ).items():
137                     for blocked in blocks:
138                         if blocked == "":
139                             continue
140                         blocked == blocked.lower()
141                         blocker == blocker.lower()
142                         c.execute(
143                             "select domain from instances where domain = ?", (blocked,)
144                         )
145                         if c.fetchone() == None:
146                             c.execute(
147                                 "insert into instances select ?, ?, ?",
148                                 (blocked, get_hash(blocked), get_type(blocked)),
149                             )
150                         c.execute(
151                             "select * from blocks where blocker = ? and blocked = ? and block_level = ?",
152                             (blocker, blocked, block_level),
153                         )
154                         if c.fetchone() == None:
155                             c.execute(
156                                 "insert into blocks select ?, ?, '', ?",
157                                 (blocker, blocked, block_level),
158                             )
159             conn.commit()
160             # Reasons
161             if "mrf_simple_info" in federation:
162                 for block_level, info in (
163                     {**federation["mrf_simple_info"],
164                     **(federation["quarantined_instances_info"]
165                     if "quarantined_instances_info" in federation
166                     else {})}
167                 ).items():
168                     for blocked, reason in info.items():
169                         blocker == blocker.lower()
170                         blocked == blocked.lower()
171                         c.execute(
172                             "update blocks set reason = ? where blocker = ? and blocked = ? and block_level = ?",
173                             (reason["reason"], blocker, blocked, block_level),
174                         )
175             conn.commit()
176         except Exception as e:
177             print("error:", e, blocker)
178     elif software == "mastodon":
179         print(blocker)
180         try:
181             json = get_mastodon_blocks(blocker)
182             for block_level, blocks in json.items():
183                 for instance in blocks:
184                     blocked, blocked_hash, reason = instance.values()
185                     blocked == blocked.lower()
186                     blocker == blocker.lower()
187                     if blocked.count("*") <= 1:
188                         c.execute(
189                             "select hash from instances where hash = ?", (blocked_hash,)
190                         )
191                         if c.fetchone() == None:
192                             c.execute(
193                                 "insert into instances select ?, ?, ?",
194                                 (blocked, get_hash(blocked), get_type(blocked)),
195                             )
196                     c.execute(
197                         "select * from blocks where blocker = ? and blocked = ? and block_level = ?",
198                         (blocker, blocked if blocked.count("*") <= 1 else blocked_hash, block_level),
199                     )
200                     if c.fetchone() == None:
201                         c.execute(
202                             "insert into blocks select ?, ?, ?, ?",
203                             (
204                                 blocker,
205                                 blocked if blocked.count("*") <= 1 else blocked_hash,
206                                 reason,
207                                 block_level,
208                             ),
209                         )
210             conn.commit()
211         except Exception as e:
212             print("error:", e, blocker)
213     elif software == "friendica":
214         print(blocker)
215         try:
216             json = get_friendica_blocks(blocker)
217             for block_level, blocks in json.items():
218                 for instance in blocks:
219                     blocked, reason = instance.values()
220                     blocked == blocked.lower()
221                     blocker == blocker.lower()
222                     c.execute(
223                         "select domain from instances where domain = ?", (blocked,)
224                     )
225                     if c.fetchone() == None:
226                         c.execute(
227                             "insert into instances select ?, ?, ?",
228                             (blocked, get_hash(blocked), get_type(blocked)),
229                         )
230                     c.execute(
231                         "select * from blocks where blocker = ? and blocked = ?",
232                         (blocker, blocked),
233                     )
234                     if c.fetchone() == None:
235                         c.execute(
236                             "insert into blocks select ?, ?, ?, ?",
237                             (
238                                 blocker,
239                                 blocked,
240                                 reason,
241                                 block_level,
242                             ),
243                         )
244             conn.commit()
245         except Exception as e:
246             print("error:", e, blocker)
247 conn.close()