1 # Fedi API Block - An aggregator for fetching blocking data from fediverse nodes
2 # Copyright (C) 2023 Free Software Foundation
4 # This program is free software: you can redistribute it and/or modify
5 # it under the terms of the GNU Affero General Public License as published
6 # by the Free Software Foundation, either version 3 of the License, or
7 # (at your option) any later version.
9 # This program is distributed in the hope that it will be useful,
10 # but WITHOUT ANY WARRANTY; without even the implied warranty of
11 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 # GNU Affero General Public License for more details.
14 # You should have received a copy of the GNU Affero General Public License
15 # along with this program. If not, see <https://www.gnu.org/licenses/>.
22 from fba.helpers import config
23 from fba.helpers import domain as domain_helper
24 from fba.helpers import tidyup
26 from fba.http import csrf
27 from fba.http import federation
28 from fba.http import network
30 from fba.models import instances
32 logging.basicConfig(level=logging.INFO)
33 logger = logging.getLogger(__name__)
34 #logger.setLevel(logging.DEBUG)
36 def fetch_peers(domain: str, origin: str) -> list:
37 logger.debug("domain='%s',origin='%s' - CALLED!", domain, origin)
38 domain_helper.raise_on(domain)
42 # No CSRF by default, you don't have to add network.api_headers by yourself here
46 logger.debug("Checking CSRF for domain='%s'", domain)
47 headers = csrf.determine(domain, dict())
48 except network.exceptions as exception:
49 logger.warning("Exception '%s' during checking CSRF (fetch_peers,%s)", type(exception), __name__)
50 instances.set_last_error(domain, exception)
52 logger.debug("Returning empty list ... - EXIT!")
56 logger.debug("Fetching '/api/v3/site' from domain='%s' ...", domain)
57 data = network.get_json_api(
61 (config.get("connection_timeout"), config.get("read_timeout"))
64 logger.debug("data[]='%s'", type(data))
65 if "error_message" in data:
66 logger.warning("Could not reach any JSON API: domain='%s'", domain)
67 instances.set_last_error(domain, data)
68 elif "federated_instances" in data["json"] and isinstance(data["json"]["federated_instances"], dict):
69 logger.debug("Found federated_instances for domain='%s'", domain)
70 peers = peers + federation.add_peers(data["json"]["federated_instances"])
72 logger.debug("Marking domain='%s' as successfully handled ...", domain)
73 instances.set_success(domain)
76 logger.warning("Fetching instances for domain='%s' from /instances ...", domain)
77 peers = fetch_instances(domain, origin)
79 except network.exceptions as exception:
80 logger.warning("Exception during fetching JSON: domain='%s',exception[%s]:'%s'", domain, type(exception), str(exception))
81 instances.set_last_error(domain, exception)
83 logger.debug("peers()=%d - EXIT!", len(peers))
86 def fetch_blocks(domain: str) -> list:
87 logger.debug("domain='%s - CALLED!", domain)
88 domain_helper.raise_on(domain)
90 if not instances.is_registered(domain):
91 raise Exception(f"domain='{domain}' is not registered but function is invoked.")
94 "Blocked Instances".lower(),
95 "Instàncies bloquejades".lower(),
96 "Blocáilte Ásc".lower(),
98 "Blokované instance".lower(),
99 "Geblokkeerde instanties".lower(),
100 "Blockerade instanser".lower(),
101 "Instàncias blocadas".lower(),
102 "Istanze bloccate".lower(),
103 "Instances bloquées".lower(),
104 "Letiltott példányok".lower(),
105 "Instancias bloqueadas".lower(),
106 "Blokeatuta dauden instantziak".lower(),
108 "Peladen Yang Diblokir".lower(),
109 "Blokerede servere".lower(),
110 "Blokitaj nodoj".lower(),
111 "Блокирани Инстанции".lower(),
112 "Blockierte Instanzen".lower(),
113 "Estetyt instanssit".lower(),
114 "Instâncias bloqueadas".lower(),
115 "Zablokowane instancje".lower(),
116 "Blokované inštancie".lower(),
117 "المثلاء المحجوبون".lower(),
118 "Užblokuoti serveriai".lower(),
119 "ブロックしたインスタンス".lower(),
120 "Блокированные Инстансы".lower(),
121 "Αποκλεισμένοι διακομιστές".lower(),
123 "Instâncias bloqueadas".lower(),
129 # json endpoint for newer mastodongs
130 logger.debug("Fetching /instances from domain='%s'", domain)
131 response = network.fetch_response(
135 (config.get("connection_timeout"), config.get("read_timeout"))
138 logger.debug("response.ok='%s',response.status_code=%d,response.text()=%d", response.ok, response.status_code, len(response.text))
139 if response.ok and response.status_code == 200 and response.text != "":
140 logger.debug("Parsing %s Bytes ...", len(response.text))
142 doc = bs4.BeautifulSoup(response.text, "html.parser")
143 logger.debug("doc[]='%s'", type(doc))
146 for criteria in [{"class": "home-instances container-lg"}, {"class": "container"}]:
147 logger.debug("criteria='%s'", criteria)
148 containers = doc.findAll("div", criteria)
150 logger.debug("Checking %d containers ...", len(containers))
151 for container in containers:
152 logger.debug("container[]='%s'", type(container))
153 for header in container.find_all(["h2", "h3", "h4", "h5"]):
155 logger.debug("header[%s]='%s' - BEFORE!", type(header), header)
156 if header is not None:
157 content = str(header.contents[0])
158 logger.debug("content[%s]='%s' - AFTER!", type(content), content)
161 logger.debug("domain='%s' has returned empty header='%s' - SKIPPED!", domain, header)
163 elif not isinstance(content, str):
164 logger.debug("content[]='%s' is not supported/wanted type 'str' - SKIPPED!", type(content))
166 elif content.lower() in translations:
167 logger.debug("Found header='%s' with blocked instances - BREAK(3) !", header)
171 logger.debug("found[]='%s'", type(found))
172 if found is not None:
173 logger.debug("Found header with blocked instances - BREAK(2) !")
176 logger.debug("found[]='%s'", type(found))
177 if found is not None:
178 logger.debug("Found header with blocked instances - BREAK(1) !")
181 logger.debug("found[]='%s'", type(found))
183 logger.info("domain='%s' has no HTML blocklist, checking scripts ...", domain)
184 peers = parse_script(doc, "blocked")
186 logger.debug("domain='%s' has %d peer(s).", domain, len(peers))
187 for blocked in peers:
188 logger.debug("Appending blocker='%s',blocked='%s',block_level='reject' ...", domain, blocked)
193 "block_level": "reject",
196 logger.debug("blocklist()=%d - EXIT!", len(blocklist))
199 blocking = found.find_next(["ul", "table"]).findAll("a")
200 logger.debug("Found %d blocked instance(s) ...", len(blocking))
202 logger.debug("tag[]='%s'", type(tag))
203 blocked = tidyup.domain(tag.contents[0]) if tag.contents[0] != "" else None
204 logger.debug("blocked='%s'", blocked)
206 if blocked is None or blocked == "":
207 logger.warning("blocked='%s' is empty after tidyup.domain() - SKIPPED!", tag.contents[0])
209 elif not domain_helper.is_wanted(blocked):
210 logger.debug("blocked='%s' is not wanted - SKIPPED!", blocked)
213 logger.debug("Appending blocker='%s',blocked='%s',block_level='reject' ...", domain, blocked)
218 "block_level": "reject",
221 logger.warning("Cannot fetch /instances due to error: response.ok='%s',response.status_code=%d,response.details='%s'", response.ok, response.status_code, response.reason)
222 instances.set_last_error(domain, response)
224 except network.exceptions as exception:
225 logger.warning("domain='%s',exception[%s]:'%s'", domain, type(exception), str(exception))
226 instances.set_last_error(domain, exception)
228 logger.debug("blocklist()=%d - EXIT!", len(blocklist))
231 def fetch_instances(domain: str, origin: str) -> list:
232 logger.debug("domain='%s',origin='%s' - CALLED!", domain, origin)
233 domain_helper.raise_on(domain)
238 # json endpoint for newer mastodongs
239 logger.debug("Fetching /instances from domain='%s'", domain)
240 response = network.fetch_response(
244 (config.get("connection_timeout"), config.get("read_timeout"))
247 logger.debug("response.ok='%s',response.status_code=%d,response.text()=%d", response.ok, response.status_code, len(response.text))
248 if response.ok and response.status_code == 200 and response.text != "":
249 logger.debug("Parsing %s Bytes ...", len(response.text))
251 doc = bs4.BeautifulSoup(response.text, "html.parser")
252 logger.debug("doc[]='%s'", type(doc))
254 for criteria in [{"class": "home-instances container-lg"}, {"class": "container"}]:
255 logger.debug("criteria='%s'", criteria)
256 containers = doc.findAll("div", criteria)
258 logger.debug("Checking %d containers ...", len(containers))
259 for header in containers:
260 logger.debug("header[%s]='%s'", type(header), header)
262 rows = header.find_next(["ul","table"]).findAll("a")
263 logger.debug("Found %d instance(s) ...", len(rows))
265 logger.debug("tag[]='%s'", type(tag))
266 text = tag.contents[0] if isinstance(tag.contents[0], str) else tag.contents[0].text
267 logger.debug("text='%s' - BEFORE!", text)
269 peer = tidyup.domain(text) if text != "" else None
270 logger.debug("peer='%s' - AFTER", peer)
272 if peer is None or peer == "":
273 logger.warning("peer='%s' is empty, text='%s' - SKIPPED!", peer, text)
275 elif not domain_helper.is_wanted(peer):
276 logger.debug("peer='%s' is not wanted - SKIPPED!", peer)
279 logger.debug("peer='%s' already added - SKIPPED!", peer)
282 logger.debug("Appending peer='%s' ...", peer)
285 logger.debug("peers()=%d", len(peers))
287 logger.debug("Found no peers for domain='%s', trying script tag ...", domain)
288 peers = parse_script(doc)
290 logger.warning("Cannot fetch /instances due to error: response.ok='%s',response.status_code=%d,response.details='%s'", response.ok, response.status_code, response.reason)
291 instances.set_last_error(domain, response)
293 logger.debug("Marking domain='%s' as successfully handled, peers()=%d ...", domain, len(peers))
294 instances.set_success(domain)
296 except network.exceptions as exception:
297 logger.warning("domain='%s',exception[%s]:'%s'", domain, type(exception), str(exception))
298 instances.set_last_error(domain, exception)
300 logger.debug("peers()=%d - EXIT!", len(peers))
303 def parse_script(doc: bs4.BeautifulSoup, only: str = None) -> list:
304 logger.debug("doc[]='%s',only='%s' - CALLED!")
306 if not isinstance(doc, bs4.BeautifulSoup):
307 raise ValueError(f"Parameter doc[]='{type(only)}' is not of type 'bs4.BeautifulSoup'")
308 elif not isinstance(only, str) and only is not None:
309 raise ValueError(f"Parameter only[]='{type(only)}' is not of type 'str'")
310 elif isinstance(only, str) and only == "":
311 raise ValueError("Parameter 'only' is empty")
313 scripts = doc.find_all("script")
316 logger.debug("scripts()=%d", len(scripts))
317 for script in scripts:
318 logger.debug("script[%s].contents()=%d", type(script), len(script.contents))
319 if len(script.contents) == 0:
320 logger.debug("script has no contents - SKIPPED!")
322 elif not script.contents[0].startswith("window.isoData"):
323 logger.debug("script.contents[0]='%s' does not start with window.isoData - SKIPPED!", script.contents[0])
326 logger.debug("script.contents[0][]='%s'", type(script.contents[0]))
328 iso_data = script.contents[0].split("=")[1].strip().replace(":undefined", ":\"undefined\"")
329 logger.debug("iso_data[%s]='%s'", type(iso_data), iso_data)
333 parsed = json.loads(iso_data)
334 except json.decoder.JSONDecodeError as exception:
335 logger.warning("Exception '%s' during parsing %d Bytes: '%s' - EXIT!", type(exception), len(iso_data), str(exception))
338 logger.debug("parsed[%s]()=%d", type(parsed), len(parsed))
340 if "routeData" not in parsed:
341 logger.warning("parsed[%s]()=%d does not contain element 'routeData'", type(parsed), len(parsed))
343 elif "federatedInstancesResponse" not in parsed["routeData"]:
344 logger.warning("parsed[routeData][%s]()=%d does not contain element 'federatedInstancesResponse'", type(parsed["routeData"]), len(parsed["routeData"]))
346 elif "data" not in parsed["routeData"]["federatedInstancesResponse"]:
347 logger.warning("parsed[routeData][federatedInstancesResponse][%s]()=%d does not contain element 'data'", type(parsed["routeData"]["federatedInstancesResponse"]), len(parsed["routeData"]["federatedInstancesResponse"]))
349 elif "federated_instances" not in parsed["routeData"]["federatedInstancesResponse"]["data"]:
350 logger.warning("parsed[routeData][federatedInstancesResponse][data][%s]()=%d does not contain element 'data'", type(parsed["routeData"]["federatedInstancesResponse"]["data"]), len(parsed["routeData"]["federatedInstancesResponse"]["data"]))
353 data = parsed["routeData"]["federatedInstancesResponse"]["data"]["federated_instances"]
354 logger.debug("Checking %d data elements ...", len(data))
356 logger.debug("element='%s'", element)
357 if isinstance(only, str) and only != element:
358 logger.debug("Skipping unwanted element='%s',only='%s'", element, only)
361 logger.debug("Checking data[%s]()=%d row(s) ...", element, len(data[element]))
362 for row in data[element]:
363 logger.debug("row[]='%s'", type(row))
364 if "domain" not in row:
365 logger.warning("row()=%d has no element 'domain' - SKIPPED!", len(row))
368 logger.debug("row[domain]='%s' - BEFORE!", row["domain"])
369 peer = tidyup.domain(row["domain"])
370 logger.debug("peer='%s' - AFTER!", peer)
372 if peer is None or peer == "":
373 logger.warning("peer='%s' is empty, row[domain]='%s' - SKIPPED!", peer, row["domain"])
375 elif not domain_helper.is_wanted(peer):
376 logger.debug("peer='%s' is not wanted - SKIPPED!", peer)
378 logger.debug("peer='%s' already added - SKIPPED!", peer)
381 logger.debug("Appending peer='%s' ...", peer)
384 logger.debug("peers()=%d - EXIT!", len(peers))