]> git.mxchange.org Git - fba.git/blob - fetch_blocks.py
Misskey support, hopefully
[fba.git] / fetch_blocks.py
1 from requests import get
2 from requests import post
3 from hashlib import sha256
4 import sqlite3
5 from bs4 import BeautifulSoup
6
7 headers = {
8     "user-agent": "Mozilla/5.0 (Windows NT 10.0; rv:102.0) Gecko/20100101 Firefox/102.0"
9 }
10
11
12 def get_mastodon_blocks(domain: str) -> dict:
13     blocks = {
14         "Suspended servers": [],
15         "Filtered media": [],
16         "Limited servers": [],
17         "Silenced servers": [],
18     }
19
20     translations = {
21         "Silenced instances": "Silenced servers",
22         "Suspended instances": "Suspended servers",
23         "Gesperrte Server": "Suspended servers",
24         "Gefilterte Medien": "Filtered media",
25         "Stummgeschaltete Server": "Silenced servers",
26         "停止済みのサーバー": "Suspended servers",
27         "メディアを拒否しているサーバー": "Filtered media",
28         "サイレンス済みのサーバー": "Silenced servers",
29         "Serveurs suspendus": "Suspended servers",
30         "Médias filtrés": "Filtered media",
31         "Serveurs limités": "Silenced servers",
32     }
33
34     try:
35         doc = BeautifulSoup(
36             get(f"https://{domain}/about/more", headers=headers, timeout=5).text,
37             "html.parser",
38         )
39     except:
40         return {}
41
42     for header in doc.find_all("h3"):
43         header_text = header.text
44         if header_text in translations:
45             header_text = translations[header_text]
46         if header_text in blocks:
47             for line in header.find_next_siblings("table")[0].find_all("tr")[1:]:
48                 blocks[header_text].append(
49                     {
50                         "domain": line.find("span").text,
51                         "hash": line.find("span")["title"][9:],
52                         "reason": line.find_all("td")[1].text.strip(),
53                     }
54                 )
55     return {
56         "reject": blocks["Suspended servers"],
57         "media_removal": blocks["Filtered media"],
58         "federated_timeline_removal": blocks["Limited servers"]
59         + blocks["Silenced servers"],
60     }
61
62 def get_friendica_blocks(domain: str) -> dict:
63     blocks = []
64
65     try:
66         doc = BeautifulSoup(
67             get(f"https://{domain}/friendica", headers=headers, timeout=5).text,
68             "html.parser",
69         )
70     except:
71         return {}
72
73     blocklist = doc.find(id="about_blocklist")
74     for line in blocklist.find("table").find_all("tr")[1:]:
75             blocks.append(
76                 {
77                     "domain": line.find_all("td")[0].text.strip(),
78                     "reason": line.find_all("td")[1].text.strip()
79                 }
80             )
81
82     return {
83         "reject": blocks
84     }
85
86 def get_pisskey_blocks(domain: str) -> dict:
87     blocks = {
88         "suspended": [],
89         "blocked": []
90     }
91
92     try:
93         counter = 0
94         step = 99
95         while True:
96             # iterating through all "suspended" (follow-only in its terminology) instances page-by-page, since that troonware doesn't support sending them all at once
97             try:
98                 if counter == 0:
99                     doc = post(f"https://{domain}/api/federation/instances", data=dumps({"sort":"+caughtAt","host":None,"suspended":True,"limit":step}), headers=headers, timeout=5).json()
100                     if doc == []: raise
101                 else:
102                     doc = post(f"https://{domain}/api/federation/instances", data=dumps({"sort":"+caughtAt","host":None,"suspended":True,"limit":step,"offset":counter-1}), headers=headers, timeout=5).json()
103                     if doc == []: raise
104                 for instance in doc:
105                     # just in case
106                     if instance["isSuspended"]:
107                         blocks["suspended"].append(
108                             {
109                                 "domain": instance["host"],
110                                 # no reason field, nothing
111                                 "reason": ""
112                             }
113                         )
114                 counter = counter + step
115             except:
116                 counter = 0
117                 break
118
119         while True:
120             # same shit, different asshole ("blocked" aka full suspend)
121             try:
122                 if counter == 0:
123                     doc = post(f"https://{domain}/api/federation/instances", data=dumps({"sort":"+caughtAt","host":None,"blocked":True,"limit":step}), headers=headers, timeout=5).json()
124                     if doc == []: raise
125                 else:
126                     doc = post(f"https://{domain}/api/federation/instances", data=dumps({"sort":"+caughtAt","host":None,"blocked":True,"limit":step,"offset":counter-1}), headers=headers, timeout=5).json()
127                     if doc == []: raise
128                 for instance in doc:
129                     if instance["isBlocked"]:
130                         blocks["blocked"].append(
131                             {
132                                 "domain": instance["host"],
133                                 "reason": ""
134                             }
135                         )
136                 counter = counter + step
137             except:
138                 counter = 0
139                 break
140
141         return {
142             "reject": blocks["blocked"],
143             "followers_only": blocks["suspended"]
144         }
145
146     except:
147         return {}
148
149 def get_hash(domain: str) -> str:
150     return sha256(domain.encode("utf-8")).hexdigest()
151
152
153 def get_type(domain: str) -> str:
154     try:
155         res = get(f"https://{domain}/nodeinfo/2.1.json", headers=headers, timeout=5)
156         if res.status_code == 404:
157             res = get(f"https://{domain}/nodeinfo/2.0", headers=headers, timeout=5)
158         if res.status_code == 404:
159             res = get(f"https://{domain}/nodeinfo/2.0.json", headers=headers, timeout=5)
160         if res.ok and "text/html" in res.headers["content-type"]:
161             res = get(f"https://{domain}/nodeinfo/2.1", headers=headers, timeout=5)
162         if res.ok:
163             if res.json()["software"]["name"] == "akkoma":
164                 return "pleroma"
165             elif res.json()["software"]["name"] == "rebased":
166                 return "pleroma"
167             elif res.json()["software"]["name"] == "hometown":
168                 return "mastodon"
169             elif res.json()["software"]["name"] == "ecko":
170                 return "mastodon"
171             elif res.json()["software"]["name"] == "calckey":
172                 return "misskey"
173             else:
174                 return res.json()["software"]["name"]
175         elif res.status_code == 404:
176             res = get(f"https://{domain}/api/v1/instance", headers=headers, timeout=5)
177         if res.ok:
178             return "mastodon"
179     except:
180         return None
181
182
183 conn = sqlite3.connect("blocks.db")
184 c = conn.cursor()
185
186 c.execute(
187     "select domain, software from instances where software in ('pleroma', 'mastodon', 'friendica', 'misskey')"
188 )
189
190 for blocker, software in c.fetchall():
191     if software == "pleroma":
192         print(blocker)
193         try:
194             # Blocks
195             federation = get(
196                 f"https://{blocker}/nodeinfo/2.1.json", headers=headers, timeout=5
197             ).json()["metadata"]["federation"]
198             if "mrf_simple" in federation:
199                 for block_level, blocks in (
200                     {**federation["mrf_simple"],
201                     **{"quarantined_instances": federation["quarantined_instances"]}}
202                 ).items():
203                     for blocked in blocks:
204                         if blocked == "":
205                             continue
206                         blocked == blocked.lower()
207                         blocker == blocker.lower()
208                         c.execute(
209                             "select domain from instances where domain = ?", (blocked,)
210                         )
211                         if c.fetchone() == None:
212                             c.execute(
213                                 "insert into instances select ?, ?, ?",
214                                 (blocked, get_hash(blocked), get_type(blocked)),
215                             )
216                         c.execute(
217                             "select * from blocks where blocker = ? and blocked = ? and block_level = ?",
218                             (blocker, blocked, block_level),
219                         )
220                         if c.fetchone() == None:
221                             c.execute(
222                                 "insert into blocks select ?, ?, '', ?",
223                                 (blocker, blocked, block_level),
224                             )
225             conn.commit()
226             # Reasons
227             if "mrf_simple_info" in federation:
228                 for block_level, info in (
229                     {**federation["mrf_simple_info"],
230                     **(federation["quarantined_instances_info"]
231                     if "quarantined_instances_info" in federation
232                     else {})}
233                 ).items():
234                     for blocked, reason in info.items():
235                         blocker == blocker.lower()
236                         blocked == blocked.lower()
237                         c.execute(
238                             "update blocks set reason = ? where blocker = ? and blocked = ? and block_level = ?",
239                             (reason["reason"], blocker, blocked, block_level),
240                         )
241             conn.commit()
242         except Exception as e:
243             print("error:", e, blocker)
244     elif software == "mastodon":
245         print(blocker)
246         try:
247             json = get_mastodon_blocks(blocker)
248             for block_level, blocks in json.items():
249                 for instance in blocks:
250                     blocked, blocked_hash, reason = instance.values()
251                     blocked == blocked.lower()
252                     blocker == blocker.lower()
253                     if blocked.count("*") <= 1:
254                         c.execute(
255                             "select hash from instances where hash = ?", (blocked_hash,)
256                         )
257                         if c.fetchone() == None:
258                             c.execute(
259                                 "insert into instances select ?, ?, ?",
260                                 (blocked, get_hash(blocked), get_type(blocked)),
261                             )
262                     c.execute(
263                         "select * from blocks where blocker = ? and blocked = ? and block_level = ?",
264                         (blocker, blocked if blocked.count("*") <= 1 else blocked_hash, block_level),
265                     )
266                     if c.fetchone() == None:
267                         c.execute(
268                             "insert into blocks select ?, ?, ?, ?",
269                             (
270                                 blocker,
271                                 blocked if blocked.count("*") <= 1 else blocked_hash,
272                                 reason,
273                                 block_level,
274                             ),
275                         )
276             conn.commit()
277         except Exception as e:
278             print("error:", e, blocker)
279     elif software == "friendica" or software == "misskey":
280         print(blocker)
281         try:
282             if software == "friendica":
283                 json = get_friendica_blocks(blocker)
284             elif software == "misskey":
285                 json = get_pisskey_blocks(blocker)
286             for block_level, blocks in json.items():
287                 for instance in blocks:
288                     blocked, reason = instance.values()
289                     blocked == blocked.lower()
290                     blocker == blocker.lower()
291                     c.execute(
292                         "select domain from instances where domain = ?", (blocked,)
293                     )
294                     if c.fetchone() == None:
295                         c.execute(
296                             "insert into instances select ?, ?, ?",
297                             (blocked, get_hash(blocked), get_type(blocked)),
298                         )
299                     c.execute(
300                         "select * from blocks where blocker = ? and blocked = ?",
301                         (blocker, blocked),
302                     )
303                     if c.fetchone() == None:
304                         c.execute(
305                             "insert into blocks select ?, ?, ?, ?",
306                             (
307                                 blocker,
308                                 blocked,
309                                 reason,
310                                 block_level,
311                             ),
312                         )
313             conn.commit()
314         except Exception as e:
315             print("error:", e, blocker)
316 conn.close()