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 import database
25 from fba.helpers import blacklist
26 from fba.helpers import config
27 from fba.helpers import domain as domain_helper
28 from fba.helpers import tidyup
30 from fba.http import network
31 from fba.http import nodeinfo
33 from fba.models import blocks
34 from fba.models import instances
36 logging.basicConfig(level=logging.INFO)
37 logger = logging.getLogger(__name__)
39 # Language mapping X -> English
42 "filtered media" : "filtered_media",
43 "limited servers" : "followers_only",
44 "followers-only" : "followers_only",
45 "media removal" : "media_removal",
46 "media_removal" : "media_removal",
47 "media force-set as sensitive": "media_nsfw",
48 "nsfw" : "media_nsfw",
50 "suspended servers": "reject",
51 "silenced servers" : "silenced",
52 "removal from \"the whole known network\" timeline": "federated_timeline_removal",
55 def fetch_blocks(domain: str) -> list:
56 logger.debug("domain='%s' - CALLED!", domain)
57 domain_helper.raise_on(domain)
59 if blacklist.is_blacklisted(domain):
60 raise Exception(f"domain='{domain}' is blacklisted but function is invoked.")
61 elif not instances.is_registered(domain):
62 raise Exception(f"domain='{domain}' is not registered but function is invoked.")
68 logger.debug("Fetching nodeinfo: domain='%s'", domain)
69 rows = nodeinfo.fetch(domain, update_mode=False)
71 if "error_message" in rows:
72 logger.warning("Error message '%s' during fetching nodeinfo for domain='%s'", rows["error_message"], domain)
73 instances.set_last_error(domain, rows)
74 instances.update(domain)
76 logger.debug("Returning empty list ... - EXIT!")
78 elif "exception" in rows:
79 logger.warning("Exception '%s' during fetching nodeinfo for domain='%s' - EXIT!", type(rows["exception"]), domain)
82 logger.debug("rows[json] found for domain='%s'", domain)
85 except network.exceptions as exception:
86 logger.warning("Exception '%s' during fetching nodeinfo from domain='%s'", type(exception), domain)
87 instances.set_last_error(domain, exception)
89 logger.debug("rows[]='%s'", type(rows))
91 logger.warning("Could not fetch nodeinfo from domain='%s' - EXIT!", domain)
93 elif "metadata" not in rows:
94 logger.warning("rows()=%d does not have key 'metadata', domain='%s' - EXIT!", len(rows), domain)
96 elif "federation" not in rows["metadata"]:
97 logger.warning("rows()=%d does not have key 'federation', domain='%s' - EXIT!", len(rows["metadata"]), domain)
101 data = rows["metadata"]["federation"]
102 logger.debug("data[]='%s'", type(data))
104 if "mrf_simple" in data:
105 logger.debug("Found mrf_simple in API response from domain='%s'", domain)
107 for block_level, blocklist in (
109 **data["mrf_simple"],
111 "quarantined_instances": data["quarantined_instances"]
115 logger.debug("block_level='%s', blocklist()=%d", block_level, len(blocklist))
116 block_level = tidyup.domain(block_level) if block_level != "" else None
117 logger.debug("block_level='%s' - AFTER!", block_level)
119 if block_level == "":
120 logger.warning("block_level is now empty!")
122 elif block_level == "accept":
123 logger.debug("domain='%s' skipping block_level='accept'", domain)
126 block_level = blocks.alias_block_level(block_level)
128 logger.debug("Checking %d entries from domain='%s',block_level='%s' ...", len(blocklist), domain, block_level)
129 for blocked in blocklist:
130 logger.debug("blocked='%s' - BEFORE!", blocked)
131 blocked = tidyup.domain(blocked) if blocked != "" else None
132 logger.debug("blocked='%s' - AFTER!", blocked)
134 if blocked in [None, ""]:
135 logger.warning("blocked='%s' is empty after tidyup.domain(): domain='%s',block_level='%s' - SKIPPED!", blocked, domain, block_level)
137 elif validators.domain(blocked) and blacklist.is_blacklisted(blocked):
138 logger.debug("blocked='%s' is blacklisted - SKIPPED!")
141 logger.debug("Invoking utils.deobfuscate(%s, %s) ...", blocked, domain)
142 blocked = utils.deobfuscate(blocked, domain)
143 logger.debug("blocked[%s]='%s' - DEOBFUSCATED!", type(blocked), blocked)
145 if blocked in [None, ""]:
146 logger.warning("instance[host]='%s' is None or empty after tidyup.domain() - SKIPPED!", instance["host"])
148 elif not domain_helper.is_wanted(blocked):
149 logger.debug("blocked='%s' is not wanted - SKIPPED!", blocked)
152 logger.debug("Appending blocker='%s',blocked='%s',block_level='%s' ...", domain, blocked, block_level)
157 "block_level": block_level,
160 elif "quarantined_instances" in data:
161 logger.debug("Found 'quarantined_instances' in JSON response: domain='%s'", domain)
163 block_level = "quarantined"
165 logger.debug("Checking %d quarantined instance(s) ...", len(data["quarantined_instances"]))
166 for blocked in data["quarantined_instances"]:
167 logger.debug("blocked='%s' - BEFORE!", blocked)
168 blocked = tidyup.domain(blocked) if blocked != "" else None
169 logger.debug("blocked='%s' - AFTER!", blocked)
171 if blocked in [None, ""]:
172 logger.warning("blocked is empty after tidyup.domain(): domain='%s',block_level='%s'", domain, block_level)
174 elif not domain_helper.is_wanted(blocked):
175 logger.debug("blocked='%s' is not wanted - SKIPPED!", blocked)
178 logger.debug("Appending blocker='%s',blocked='%s',block_level='%s' ...", domain, blocked, block_level)
183 "block_level": block_level,
187 logger.warning("Cannot find 'mrf_simple' or 'quarantined_instances' in JSON reply: domain='%s'", domain)
189 logger.debug("Invoking commit() ...")
190 database.connection.commit()
193 if "mrf_simple_info" in data:
194 logger.debug("Found mrf_simple_info in API response: domain='%s'", domain)
196 for block_level, info in (
198 **data["mrf_simple_info"],
199 **(data["quarantined_instances_info"] if "quarantined_instances_info" in data else {})
202 logger.debug("block_level='%s', info.items()=%d", block_level, len(info.items()))
203 block_level = tidyup.domain(block_level) if block_level != "" else None
204 logger.debug("block_level='%s' - AFTER!", block_level)
206 if block_level in [None, ""]:
207 logger.warning("block_level='%s' is now empty!", block_level)
209 elif block_level == "accept":
210 logger.debug("domain='%s': Skipping block_level='%s' ...", domain, block_level)
213 block_level = blocks.alias_block_level(block_level)
215 logger.debug("Checking %d entries from domain='%s',block_level='%s' ...", len(info.items()), domain, block_level)
216 for blocked, reason in info.items():
217 logger.debug("blocked='%s',reason[%s]='%s' - BEFORE!", blocked, type(reason), reason)
218 blocked = tidyup.domain(blocked) if blocked != "" else None
219 logger.debug("blocked='%s' - AFTER!", blocked)
221 if isinstance(reason, str):
222 logger.debug("reason[] is a string")
223 reason = tidyup.reason(reason)
224 elif isinstance(reason, dict) and "reason" in reason:
225 logger.debug("reason[] is a dict")
226 reason = tidyup.reason(reason["reason"]) if isinstance(reason["reason"], str) else None
227 elif reason is not None:
228 raise ValueError(f"Cannot handle reason[]='{type(reason)}'")
230 logger.debug("blocked='%s',reason='%s' - AFTER!", blocked, reason)
233 logger.warning("blocked is empty after tidyup.domain(): domain='%s',block_level='%s'", domain, block_level)
236 logger.debug("Checking %d blockdict records ...", len(blockdict))
237 for block in blockdict:
238 logger.debug("block[blocked]='%s',blocked='%s'", block["blocked"], blocked)
239 if block["blocked"] == blocked:
240 logger.debug("Updating reason='%s' for blocker='%s'", reason, block["blocked"])
241 block["reason"] = reason
243 elif "quarantined_instances_info" in data and "quarantined_instances" in data["quarantined_instances_info"]:
244 logger.debug("Found 'quarantined_instances_info' in JSON response: domain='%s'", domain)
246 block_level = "quarantined"
248 #print(data["quarantined_instances_info"])
249 rows = data["quarantined_instances_info"]["quarantined_instances"]
251 logger.debug("blocked='%s' - BEFORE!", blocked)
252 reason = tidyup.reason(rows[blocked]["reason"]) if rows[blocked]["reason"] != "" else None
253 blocked = tidyup.domain(blocked) if blocked != "" else None
254 logger.debug("blocked='%s',reason='%s' - AFTER!", blocked, reason)
256 if blocked not in rows or "reason" not in rows[blocked]:
257 logger.warning("Cannot find blocked='%s' in rows()=%d,domain='%s' - BREAK!", blocked, len(rows), domain)
260 logger.warning("blocked is empty after tidyup.domain(): domain='%s',block_level='%s'", domain, block_level)
263 logger.debug("Checking %d blockdict record(s) ...", len(blockdict))
264 for block in blockdict:
265 logger.debug("block[blocked]='%s',blocked='%s'", block["blocked"], blocked)
266 if block["blocked"] == blocked:
267 logger.debug("Updating reason='%s' for blocker='%s'", reason, block["blocked"])
268 block["reason"] = reason
270 logger.warning("Cannot find 'mrf_simple_info' or 'quarantined_instances_info' in JSON reply: domain='%s'", domain)
273 logger.debug("Did not find any useable JSON elements, domain='%s', continuing with /about page ...", domain)
274 blocklist = fetch_blocks_from_about(domain)
276 logger.debug("blocklist()=%d", len(blocklist))
277 if len(blocklist) > 0:
278 logger.info("Checking %d different blocklist(s) ...", len(blocklist))
279 for block_level in blocklist:
280 logger.debug("Checking blocklist[%s]()=%d entries ...", block_level, blocklist[block_level])
281 for block in blocklist[block_level]:
282 logger.debug("Appending blocker='%s',block[blocked]='%s',block[reason]='%s',block_level='%s' ...",domain, block["blocked"], block["reason"], block_level)
285 "blocked" : block["blocked"],
286 "reason" : block["reason"],
287 "block_level": block_level,
290 logger.debug("blockdict()=%d - EXIT!", len(blockdict))
293 def fetch_blocks_from_about(domain: str) -> dict:
294 logger.debug("domain='%s' - CALLED!", domain)
295 domain_helper.raise_on(domain)
297 if blacklist.is_blacklisted(domain):
298 raise Exception(f"domain='{domain}' is blacklisted but function is invoked.")
299 elif not instances.is_registered(domain):
300 raise Exception(f"domain='{domain}' is not registered but function is invoked.")
302 logger.debug("Fetching mastodon blocks from domain='%s'", domain)
304 for path in ["/instance/about/index.html"]:
309 logger.debug("Fetching path='%s' from domain='%s' ...", path, domain)
310 response = network.fetch_response(
314 (config.get("connection_timeout"), config.get("read_timeout"))
317 logger.debug("response.ok='%s',response.status_code=%d,response.text()=%d", response.ok, response.status_code, len(response.text))
318 if not response.ok or response.text.strip() == "":
319 logger.warning("path='%s' does not exist on domain='%s' - SKIPPED!", path, domain)
322 logger.debug("Parsing response.text()=%d Bytes ...", len(response.text))
323 doc = bs4.BeautifulSoup(
328 logger.debug("doc[]='%s'", type(doc))
329 if doc.find("h2") is not None:
330 logger.debug("Found 'h2' header in path='%s' - BREAK!", path)
333 except network.exceptions as exception:
334 logger.warning("Cannot fetch from domain='%s',exception[%s]='%s'", domain, type(exception), str(exception))
335 instances.set_last_error(domain, exception)
340 "filtered_media": [],
341 "followers_only": [],
344 "media_removal" : [],
345 "federated_timeline_removal": [],
348 logger.debug("doc[]='%s'", type(doc))
350 logger.warning("Cannot fetch any /about pages for domain='%s' - EXIT!", domain)
353 headers = doc.find_all("h2")
355 logger.debug("headers[]='%s'", type(headers))
357 logger.warning("Cannot fetch any /about pages for domain='%s' - EXIT!", domain)
360 logger.info("Checking %d headers ...", len(headers))
361 for header in headers:
362 logger.debug("header[%s]='%s'", type(header), header)
363 block_level = tidyup.reason(header.text).lower()
365 logger.debug("block_level='%s' - BEFORE!", block_level)
366 if block_level in language_mapping:
367 logger.debug("block_level='%s' - FOUND!", block_level)
368 block_level = language_mapping[block_level].lower()
370 logger.warning("block_level='%s' not found in language mapping table", block_level)
372 logger.debug("block_level='%s - AFTER!'", block_level)
373 if block_level in blocklist:
374 # replaced find_next_siblings with find_all_next to account for instances that e.g. hide lists in dropdown menu
375 logger.debug("Found block_level='%s', importing domain blocks ...", block_level)
376 for line in header.find_next("table").find_all("tr")[1:]:
377 logger.debug("line[]='%s'", type(line))
378 blocked = line.find_all("td")[0].text
379 reason = line.find_all("td")[1].text
381 logger.debug("blocked='%s',reason='%s'", blocked, reason)
382 blocked = tidyup.domain(blocked) if blocked != "" else None
383 reason = tidyup.reason(reason) if reason != "" else None
384 logger.debug("blocked='%s',reason='%s' - AFTER!", blocked, reason)
386 if blocked in [None, ""]:
387 logger.debug("domain='%s',block_level='%s': blocked='%s' is empty - SKIPPED!", domain, block_level, blocked)
389 elif not domain_helper.is_wanted(blocked):
390 logger.debug("blocked='%s' is not wanted - SKIPPED!", blocked)
393 logger.debug("Appending block_level='%s',blocked='%s',reason='%s' ...", block_level, blocked, reason)
394 blocklist[block_level].append({
399 logger.warning("block_level='%s' not found in blocklist()=%d", block_level, len(blocklist))
401 logger.debug("Returning blocklist for domain='%s' - EXIT!", domain)