fba/commands.py

   1 # Fedi API Block - An aggregator for fetching blocking data from fediverse nodes
   2 # Copyright (C) 2023 Free Software Foundation
   3 #
   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.
   8 #
   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.
  13 #
  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/>.
  16
  17 import csv
  18 import inspect
  19 import json
  20 import logging
  21 import time
  22
  23 import argparse
  24 import atoma
  25 import bs4
  26 import markdown
  27 import reqto
  28 import validators
  29
  30 from fba import csrf
  31 from fba import database
  32 from fba import utils
  33
  34 from fba.helpers import blacklist
  35 from fba.helpers import config
  36 from fba.helpers import cookies
  37 from fba.helpers import locking
  38 from fba.helpers import tidyup
  39
  40 from fba.http import federation
  41 from fba.http import network
  42
  43 from fba.models import blocks
  44 from fba.models import instances
  45
  46 from fba.networks import friendica
  47 from fba.networks import lemmy
  48 from fba.networks import mastodon
  49 from fba.networks import misskey
  50 from fba.networks import pleroma
  51
  52 logging.basicConfig(level=logging.INFO)
  53 logger = logging.getLogger(__name__)
  54 #logger.setLevel(logging.DEBUG)
  55
  56 def check_instance(args: argparse.Namespace) -> int:
  57     logger.debug("args.domain='%s' - CALLED!", args.domain)
  58     status = 0
  59     if not validators.domain(args.domain):
  60         logger.warning("args.domain='%s' is not valid", args.domain)
  61         status = 100
  62     elif blacklist.is_blacklisted(args.domain):
  63         logger.warning("args.domain='%s' is blacklisted", args.domain)
  64         status = 101
  65     elif instances.is_registered(args.domain):
  66         logger.warning("args.domain='%s' is already registered", args.domain)
  67         status = 102
  68     else:
  69         logger.info("args.domain='%s' is not known", args.domain)
  70
  71     logger.debug("status='%d' - EXIT!", status)
  72     return status
  73
  74 def fetch_pixelfed_api(args: argparse.Namespace) -> int:
  75     logger.debug("args[]='%s' - CALLED!", type(args))
  76
  77     # No CSRF by default, you don't have to add network.api_headers by yourself here
  78     headers = tuple()
  79
  80     try:
  81         logger.debug("Checking CSRF from pixelfed.org")
  82         headers = csrf.determine("pixelfed.org", dict())
  83     except network.exceptions as exception:
  84         logger.warning("Exception '%s' during checking CSRF (fetch_peers,%s) - EXIT!", type(exception), __name__)
  85         return list()
  86
  87     domains = list()
  88     try:
  89         logger.debug("Fetching JSON from pixelfed.org API, headers()=%d ...", len(headers))
  90         fetched = network.get_json_api(
  91             "pixelfed.org",
  92             "/api/v1/servers/all.json?scope=All&country=all&language=all",
  93             headers,
  94             (config.get("connection_timeout"), config.get("read_timeout"))
  95         )
  96
  97         logger.debug("JSON API returned %d elements", len(fetched))
  98         if "error_message" in fetched:
  99             logger.warning("API returned error_message='%s' - EXIT!", fetched["error_message"])
 100             return 101
 101         elif "data" not in fetched["json"]:
 102             logger.warning("API did not return JSON with 'data' element - EXIT!")
 103             return 102
 104
 105         rows = fetched["json"]["data"]
 106         logger.info("Checking %d fetched rows ...", len(rows))
 107         for row in rows:
 108             logger.debug("row[]='%s'", type(row))
 109             if "domain" not in row:
 110                 logger.warning("row='%s' does not contain element 'domain' - SKIPPED!")
 111                 continue
 112             elif not utils.is_domain_wanted(row['domain']):
 113                 logger.debug("row[domain]='%s' is not wanted - SKIPPED!", row['domain'])
 114                 continue
 115             elif instances.is_registered(row['domain']):
 116                 logger.debug("row[domain]='%s' is already registered - SKIPPED!", row['domain'])
 117                 continue
 118
 119             logger.debug("Fetching instances from row[domain]='%s' ...", row['domain'])
 120             federation.fetch_instances(row['domain'], None, None, inspect.currentframe().f_code.co_name)
 121
 122     except network.exceptions as exception:
 123         logger.warning("Cannot fetch graphql,exception[%s]:'%s' - EXIT!", type(exception), str(exception))
 124         return 103
 125
 126     logger.debug("Success! - EXIT!")
 127     return 0
 128
 129 def fetch_bkali(args: argparse.Namespace) -> int:
 130     logger.debug("args[]='%s' - CALLED!", type(args))
 131     domains = list()
 132     try:
 133         fetched = network.post_json_api("gql.api.bka.li", "/v1/graphql", json.dumps({
 134             "query": "query domainlist {nodeinfo(order_by: {domain: asc}) {domain}}"
 135         }))
 136
 137         logger.debug("fetched[]='%s'", type(fetched))
 138         if "error_message" in fetched:
 139             logger.warning("post_json_api() for 'gql.api.bka.li' returned error message='%s", fetched['error_message'])
 140             return 100
 141         elif isinstance(fetched["json"], dict) and "error" in fetched["json"] and "message" in fetched["json"]["error"]:
 142             logger.warning("post_json_api() returned error: '%s", fetched['error']['message'])
 143             return 101
 144
 145         rows = fetched["json"]
 146
 147         logger.debug("rows(%d)[]='%s'", len(rows), type(rows))
 148         if len(rows) == 0:
 149             raise Exception("WARNING: Returned no records")
 150         elif "data" not in rows:
 151             raise Exception(f"WARNING: rows()={len(rows)} does not contain key 'data'")
 152         elif "nodeinfo" not in rows["data"]:
 153             raise Exception(f"WARNING: rows()={len(rows['data'])} does not contain key 'nodeinfo'")
 154
 155         for entry in rows["data"]["nodeinfo"]:
 156             logger.debug("entry[%s]='%s'", type(entry), entry)
 157             if "domain" not in entry:
 158                 logger.warning("entry()=%d does not contain 'domain' - SKIPPED!", len(entry))
 159                 continue
 160             elif not utils.is_domain_wanted(entry["domain"]):
 161                 logger.debug("entry[domain]='%s' is not wanted - SKIPPED!")
 162                 continue
 163             elif instances.is_registered(entry["domain"]):
 164                 logger.debug("domain='%s' is already registered - SKIPPED!", entry['domain'])
 165                 continue
 166
 167             logger.debug("Adding domain='%s' ...", entry['domain'])
 168             domains.append(entry["domain"])
 169
 170     except network.exceptions as exception:
 171         logger.warning("Cannot fetch graphql,exception[%s]:'%s' - EXIT!", type(exception), str(exception))
 172         return 102
 173
 174     logger.debug("domains()=%d", len(domains))
 175     if len(domains) > 0:
 176         locking.acquire()
 177
 178         logger.info("Adding %d new instances ...", len(domains))
 179         for domain in domains:
 180             try:
 181                 logger.info("Fetching instances from domain='%s' ...", domain)
 182                 federation.fetch_instances(domain, 'tak.teleyal.blog', None, inspect.currentframe().f_code.co_name)
 183
 184                 logger.debug("Invoking cookies.clear(%s) ...", domain)
 185                 cookies.clear(domain)
 186             except network.exceptions as exception:
 187                 logger.warning("Exception '%s' during fetching instances (fetch_bkali) from domain='%s'", type(exception), domain)
 188                 instances.set_last_error(domain, exception)
 189
 190     logger.debug("Success - EXIT!")
 191     return 0
 192
 193 def fetch_blocks(args: argparse.Namespace) -> int:
 194     logger.debug("args[]='%s' - CALLED!", type(args))
 195     if args.domain is not None and args.domain != "":
 196         logger.debug("args.domain='%s' - checking ...", args.domain)
 197         if not validators.domain(args.domain):
 198             logger.warning("args.domain='%s' is not valid.", args.domain)
 199             return 100
 200         elif blacklist.is_blacklisted(args.domain):
 201             logger.warning("args.domain='%s' is blacklisted, won't check it!", args.domain)
 202             return 101
 203         elif not instances.is_registered(args.domain):
 204             logger.warning("args.domain='%s' is not registered, please run ./utils.py fetch_instances '%s' first.", args.domain, args.domain)
 205             return 102
 206
 207     locking.acquire()
 208
 209     if args.domain is not None and args.domain != "":
 210         # Re-check single domain
 211         logger.debug("Querying database for single args.domain='%s' ...", args.domain)
 212         database.cursor.execute(
 213             "SELECT domain, software, origin, nodeinfo_url FROM instances WHERE domain = ?", [args.domain]
 214         )
 215     elif args.software is not None and args.software != "":
 216         # Re-check single software
 217         logger.debug("Querying database for args.software='%s' ...", args.software)
 218         database.cursor.execute(
 219             "SELECT domain, software, origin, nodeinfo_url FROM instances WHERE software = ?", [args.software]
 220         )
 221     else:
 222         # Re-check after "timeout" (aka. minimum interval)
 223         database.cursor.execute(
 224             "SELECT domain, software, origin, nodeinfo_url FROM instances WHERE software IN ('pleroma', 'mastodon', 'lemmy', 'friendica', 'misskey', 'peertube') AND (last_blocked IS NULL OR last_blocked < ?) ORDER BY rowid DESC", [time.time() - config.get("recheck_block")]
 225         )
 226
 227     rows = database.cursor.fetchall()
 228     logger.info("Checking %d entries ...", len(rows))
 229     for blocker, software, origin, nodeinfo_url in rows:
 230         logger.debug("blocker='%s',software='%s',origin='%s',nodeinfo_url='%s'", blocker, software, origin, nodeinfo_url)
 231         blocker = tidyup.domain(blocker)
 232         logger.debug("blocker='%s' - AFTER!", blocker)
 233
 234         if blocker == "":
 235             logger.warning("blocker is now empty!")
 236             continue
 237         elif nodeinfo_url is None or nodeinfo_url == "":
 238             logger.debug("blocker='%s',software='%s' has empty nodeinfo_url", blocker, software)
 239             continue
 240         elif not utils.is_domain_wanted(blocker):
 241             logger.warning("blocker='%s' is not wanted - SKIPPED!", blocker)
 242             continue
 243
 244         logger.debug("blocker='%s'", blocker)
 245         instances.set_last_blocked(blocker)
 246         instances.set_has_obfuscation(blocker, False)
 247
 248         blocking = list()
 249         blockdict = list()
 250         if software == "pleroma":
 251             logger.info("blocker='%s',software='%s'", blocker, software)
 252             blocking = pleroma.fetch_blocks(blocker, nodeinfo_url)
 253         elif software == "mastodon":
 254             logger.info("blocker='%s',software='%s'", blocker, software)
 255             blocking = mastodon.fetch_blocks(blocker, nodeinfo_url)
 256         elif software == "lemmy":
 257             logger.info("blocker='%s',software='%s'", blocker, software)
 258             blocking = lemmy.fetch_blocks(blocker, nodeinfo_url)
 259         elif software == "friendica":
 260             logger.info("blocker='%s',software='%s'", blocker, software)
 261             blocking = friendica.fetch_blocks(blocker)
 262         elif software == "misskey":
 263             logger.info("blocker='%s',software='%s'", blocker, software)
 264             blocking = misskey.fetch_blocks(blocker)
 265         else:
 266             logger.warning("Unknown software: blocker='%s',software='%s'", blocker, software)
 267
 268         logger.info("Checking %d entries from blocker='%s',software='%s' ...", len(blocking), blocker, software)
 269         for block in blocking:
 270             logger.debug("blocked='%s',block_level='%s',reason='%s'", block['blocked'], block['block_level'], block['reason'])
 271
 272             if block['block_level'] == "":
 273                 logger.warning("block_level is empty, blocker='%s',blocked='%s'", block['blocker'], block['blocked'])
 274                 continue
 275
 276             logger.debug("blocked='%s',reason='%s' - BEFORE!", block['blocked'], block['reason'])
 277             block['blocked'] = tidyup.domain(block['blocked'])
 278             block['reason']  = tidyup.reason(block['reason']) if block['reason'] is not None and block['reason'] != "" else None
 279             logger.debug("blocked='%s',reason='%s' - AFTER!", block['blocked'], block['reason'])
 280
 281             if block['blocked'] == "":
 282                 logger.warning("blocked is empty, blocker='%s'", blocker)
 283                 continue
 284             elif block['blocked'].count("*") > 0:
 285                 logger.debug("blocker='%s' uses obfuscated domains, marking ...", blocker)
 286                 instances.set_has_obfuscation(blocker, True)
 287
 288                 # Some friendica servers also obscure domains without hash
 289                 row = instances.deobfuscate("*", block['blocked'])
 290
 291                 logger.debug("row[]='%s'", type(row))
 292                 if row is None:
 293                     logger.warning("Cannot deobfuscate blocked='%s',blocker='%s',software='%s' - SKIPPED!", block['blocked'], blocker, software)
 294                     continue
 295
 296                 block['blocked'] = row[0]
 297                 origin           = row[1]
 298                 nodeinfo_url     = row[2]
 299             elif block['blocked'].count("?") > 0:
 300                 logger.debug("blocker='%s' uses obfuscated domains, marking ...", blocker)
 301                 instances.set_has_obfuscation(blocker, True)
 302
 303                 # Some obscure them with question marks, not sure if that's dependent on version or not
 304                 row = instances.deobfuscate("?", block['blocked'])
 305
 306                 logger.debug("row[]='%s'", type(row))
 307                 if row is None:
 308                     logger.warning("Cannot deobfuscate blocked='%s',blocker='%s',software='%s' - SKIPPED!", block['blocked'], blocker, software)
 309                     continue
 310
 311                 block['blocked'] = row[0]
 312                 origin           = row[1]
 313                 nodeinfo_url     = row[2]
 314
 315             logger.debug("Looking up instance by domainm, blocked='%s'", block['blocked'])
 316             if not utils.is_domain_wanted(block['blocked']):
 317                 logger.debug("blocked='%s' is not wanted - SKIPPED!", block['blocked'])
 318                 continue
 319             elif block['block_level'] in ["accept", "accepted"]:
 320                 logger.debug("blocked='%s' is accepted, not wanted here - SKIPPED!", block['blocked'])
 321                 continue
 322             elif not instances.is_registered(block['blocked']):
 323                 logger.debug("Hash wasn't found, adding: blocked='%s',blocker='%s'", block['blocked'], blocker)
 324                 federation.fetch_instances(block['blocked'], blocker, None, inspect.currentframe().f_code.co_name)
 325
 326             if block['block_level'] == "silence":
 327                 logger.debug("Block level 'silence' has been changed to 'silenced'")
 328                 block['block_level'] = "silenced"
 329             elif block['block_level'] == "suspend":
 330                 logger.debug("Block level 'suspend' has been changed to 'suspended'")
 331                 block['block_level'] = "suspended"
 332
 333             if not blocks.is_instance_blocked(blocker, block['blocked'], block['block_level']):
 334                 logger.debug("Invoking blocks.add_instance(%s, %s, %s, %s)", blocker, block['blocked'], block['reason'], block['block_level'])
 335                 blocks.add_instance(blocker, block['blocked'], block['reason'], block['block_level'])
 336
 337                 logger.debug("block_level='%s',config[bot_enabled]=%s", block['block_level'], config.get("bot_enabled"))
 338                 if block['block_level'] == "reject" and config.get("bot_enabled"):
 339                     logger.debug("blocker='%s' has blocked '%s' with reason='%s' - Adding to bot notification ...", blocker, block['blocked'], block['reason'])
 340                     blockdict.append({
 341                         "blocked": block['blocked'],
 342                         "reason" : block['reason'],
 343                     })
 344             else:
 345                 logger.debug("Updating block last seen and reason for blocker='%s',blocked='%s' ...", blocker, block['blocked'])
 346                 blocks.update_last_seen(blocker, block['blocked'], block['block_level'])
 347                 blocks.update_reason(block['reason'], blocker, block['blocked'], block['block_level'])
 348
 349             logger.debug("Invoking cookies.clear(%s) ...", block['blocked'])
 350             cookies.clear(block['blocked'])
 351
 352         if instances.has_pending(blocker):
 353             logger.debug("Invoking instances.update_data(%s) ...", blocker)
 354             instances.update_data(blocker)
 355
 356         logger.debug("Invoking commit() ...")
 357         database.connection.commit()
 358
 359         logger.debug("Invoking cookies.clear(%s) ...", blocker)
 360         cookies.clear(blocker)
 361
 362         logger.debug("config[bot_enabled]='%s',blockdict()=%d'", config.get("bot_enabled"), len(blockdict))
 363         if config.get("bot_enabled") and len(blockdict) > 0:
 364             logger.info("Sending bot POST for blocker='%s,blockdict()=%d ...", blocker, len(blockdict))
 365             network.send_bot_post(blocker, blockdict)
 366
 367     logger.debug("Success! - EXIT!")
 368     return 0
 369
 370 def fetch_observer(args: argparse.Namespace) -> int:
 371     logger.debug("args[]='%s' - CALLED!", type(args))
 372     types = [
 373         "akoma",
 374         "birdsitelive",
 375         "bookwyrm",
 376         "calckey",
 377         "diaspora",
 378         "foundkey",
 379         "friendica",
 380         "funkwhale",
 381         "gancio",
 382         "gnusocial",
 383         "gotosocial",
 384         "hometown",
 385         "hubzilla",
 386         "kbin",
 387         "ktistec",
 388         "lemmy",
 389         "mastodon",
 390         "microblogpub",
 391         "misskey",
 392         "mitra",
 393         "mobilizon",
 394         "owncast",
 395         "peertube",
 396         "pixelfed",
 397         "pleroma",
 398         "plume",
 399         "snac",
 400         "takahe",
 401         "wildebeest",
 402         "writefreely"
 403     ]
 404
 405     locking.acquire()
 406
 407     logger.info("Fetching %d different table data ...", len(types))
 408     for software in types:
 409         doc = None
 410
 411         try:
 412             logger.debug("Fetching table data for software='%s' ...", software)
 413             raw = utils.fetch_url(
 414                 f"https://fediverse.observer/app/views/tabledata.php?software={software}",
 415                 network.web_headers,
 416                 (config.get("connection_timeout"), config.get("read_timeout"))
 417             ).text
 418             logger.debug("raw[%s]()=%d", type(raw), len(raw))
 419
 420             doc = bs4.BeautifulSoup(raw, features='html.parser')
 421             logger.debug("doc[]='%s'", type(doc))
 422         except network.exceptions as exception:
 423             logger.warning("Cannot fetch software='%s' from fediverse.observer: '%s'", software, type(exception))
 424             continue
 425
 426         items = doc.findAll("a", {"class": "url"})
 427         logger.info("Checking %d items,software='%s' ...", len(items), software)
 428         for item in items:
 429             logger.debug("item[]='%s'", type(item))
 430             domain = item.decode_contents()
 431
 432             logger.debug("domain='%s'", domain)
 433             if not utils.is_domain_wanted(domain):
 434                 logger.debug("domain='%s' is not wanted - SKIPPED!", domain)
 435                 continue
 436             elif instances.is_registered(domain):
 437                 logger.debug("domain='%s' is already registered - SKIPPED!", domain)
 438                 continue
 439
 440             logger.info("Fetching instances for domain='%s',software='%s'", domain, software)
 441             federation.fetch_instances(domain, None, None, inspect.currentframe().f_code.co_name)
 442
 443             logger.debug("Invoking cookies.clear(%s) ...", domain)
 444             cookies.clear(domain)
 445
 446     logger.debug("Success! - EXIT!")
 447     return 0
 448
 449 def fetch_todon_wiki(args: argparse.Namespace) -> int:
 450     logger.debug("args[]='%s' - CALLED!", type(args))
 451
 452     locking.acquire()
 453     blocklist = {
 454         "silenced": list(),
 455         "reject": list(),
 456     }
 457
 458     raw = utils.fetch_url("https://wiki.todon.eu/todon/domainblocks", network.web_headers, (config.get("connection_timeout"), config.get("read_timeout"))).text
 459     logger.debug("raw()=%d,raw[]='%s'", len(raw), type(raw))
 460
 461     doc = bs4.BeautifulSoup(raw, "html.parser")
 462     logger.debug("doc[]='%s'", type(doc))
 463
 464     silenced = doc.find("h3", {"id": "silencedlimited_servers"}).find_next("ul").findAll("li")
 465     logger.info("Checking %d silenced/limited entries ...", len(silenced))
 466     blocklist["silenced"] = utils.find_domains(silenced, "div")
 467
 468     suspended = doc.find("h3", {"id": "suspended_servers"}).find_next("ul").findAll("li")
 469     logger.info("Checking %d suspended entries ...", len(suspended))
 470     blocklist["reject"] = utils.find_domains(suspended, "div")
 471
 472     for block_level in blocklist:
 473         blockers = blocklist[block_level]
 474
 475         logger.debug("block_level='%s',blockers()=%d'", block_level, len(blockers))
 476         for blocked in blockers:
 477             logger.debug("blocked='%s'", blocked)
 478
 479             if not instances.is_registered(blocked):
 480                 try:
 481                     logger.info("Fetching instances from domain='%s' ...", blocked)
 482                     federation.fetch_instances(blocked, 'chaos.social', None, inspect.currentframe().f_code.co_name)
 483
 484                     logger.debug("Invoking cookies.clear(%s) ...", blocked)
 485                     cookies.clear(blocked)
 486                 except network.exceptions as exception:
 487                     logger.warning("Exception '%s' during fetching instances (fetch_cs) from blocked='%s'", type(exception), blocked)
 488                     instances.set_last_error(blocked, exception)
 489
 490             if blocks.is_instance_blocked("todon.eu", blocked, block_level):
 491                 logger.debug("blocked='%s',block_level='%s' is already blocked - SKIPPED!", blocked, block_level)
 492                 continue
 493
 494             logger.info("Adding new block: blocked='%s',block_level='%s'", blocked, block_level)
 495             blocks.add_instance("todon.eu", blocked, None, block_level)
 496
 497         logger.debug("Invoking commit() ...")
 498         database.connection.commit()
 499
 500     logger.debug("Success! - EXIT!")
 501     return 0
 502
 503 def fetch_cs(args: argparse.Namespace):
 504     logger.debug("args[]='%s' - CALLED!", type(args))
 505     extensions = [
 506         "extra",
 507         "abbr",
 508         "attr_list",
 509         "def_list",
 510         "fenced_code",
 511         "footnotes",
 512         "md_in_html",
 513         "admonition",
 514         "codehilite",
 515         "legacy_attrs",
 516         "legacy_em",
 517         "meta",
 518         "nl2br",
 519         "sane_lists",
 520         "smarty",
 521         "toc",
 522         "wikilinks"
 523     ]
 524
 525     domains = {
 526         "silenced": list(),
 527         "reject"  : list(),
 528     }
 529
 530     raw = utils.fetch_url("https://raw.githubusercontent.com/chaossocial/meta/master/federation.md", network.web_headers, (config.get("connection_timeout"), config.get("read_timeout"))).text
 531     logger.debug("raw()=%d,raw[]='%s'", len(raw), type(raw))
 532
 533     doc = bs4.BeautifulSoup(markdown.markdown(raw, extensions=extensions), features='html.parser')
 534     logger.debug("doc()=%d[]='%s'", len(doc), type(doc))
 535
 536     silenced = doc.find("h2", {"id": "silenced-instances"}).findNext("table").find("tbody")
 537     logger.debug("silenced[%s]()=%d", type(silenced), len(silenced))
 538     domains["silenced"] = federation.find_domains(silenced)
 539
 540     blocked = doc.find("h2", {"id": "blocked-instances"}).findNext("table").find("tbody")
 541     logger.debug("blocked[%s]()=%d", type(blocked), len(blocked))
 542     domains["reject"] = federation.find_domains(blocked)
 543
 544     logger.debug("domains[silenced]()=%d,domains[reject]()=%d", len(domains["silenced"]), len(domains["reject"]))
 545     if len(domains) > 0:
 546         locking.acquire()
 547
 548         for block_level in domains:
 549             logger.info("block_level='%s' has %d row(s)", block_level, len(domains[block_level]))
 550
 551             for row in domains[block_level]:
 552                 logger.debug("row[%s]='%s'", type(row), row)
 553                 if not instances.is_registered(row["domain"]):
 554                     try:
 555                         logger.info("Fetching instances from domain='%s' ...", row["domain"])
 556                         federation.fetch_instances(row["domain"], 'chaos.social', None, inspect.currentframe().f_code.co_name)
 557
 558                         logger.debug("Invoking cookies.clear(%s) ...", row["domain"])
 559                         cookies.clear(row["domain"])
 560                     except network.exceptions as exception:
 561                         logger.warning("Exception '%s' during fetching instances (fetch_cs) from row[domain]='%s'", type(exception), row["domain"])
 562                         instances.set_last_error(row["domain"], exception)
 563
 564                 if not blocks.is_instance_blocked('chaos.social', row["domain"], block_level):
 565                     logger.debug("domain='%s',block_level='%s' blocked by chaos.social, adding ...", row["domain"], block_level)
 566                     blocks.add_instance('chaos.social', row["domain"], row["reason"], block_level)
 567
 568         logger.debug("Invoking commit() ...")
 569         database.connection.commit()
 570
 571     logger.debug("Success! - EXIT!")
 572     return 0
 573
 574 def fetch_fba_rss(args: argparse.Namespace) -> int:
 575     logger.debug("args[]='%s' - CALLED!", type(args))
 576     domains = list()
 577
 578     logger.info("Fetch FBA-specific RSS args.feed='%s' ...", args.feed)
 579     response = utils.fetch_url(args.feed, network.web_headers, (config.get("connection_timeout"), config.get("read_timeout")))
 580
 581     logger.debug("response.ok='%s',response.status_code=%d,response.text()=%d", response.ok, response.status_code, len(response.text))
 582     if response.ok and response.status_code < 300 and len(response.text) > 0:
 583         logger.debug("Parsing RSS feed (%d Bytes) ...", len(response.text))
 584         rss = atoma.parse_rss_bytes(response.content)
 585
 586         logger.debug("rss[]='%s'", type(rss))
 587         for item in rss.items:
 588             logger.debug("item='%s'", item)
 589             domain = item.link.split("=")[1]
 590
 591             if blacklist.is_blacklisted(domain):
 592                 logger.debug("domain='%s' is blacklisted - SKIPPED!", domain)
 593                 continue
 594             elif domain in domains:
 595                 logger.debug("domain='%s' is already added - SKIPPED!", domain)
 596                 continue
 597             elif instances.is_registered(domain):
 598                 logger.debug("domain='%s' is already registered - SKIPPED!", domain)
 599                 continue
 600
 601             logger.debug("Adding domain='%s'", domain)
 602             domains.append(domain)
 603
 604     logger.debug("domains()=%d", len(domains))
 605     if len(domains) > 0:
 606         locking.acquire()
 607
 608         logger.info("Adding %d new instances ...", len(domains))
 609         for domain in domains:
 610             try:
 611                 logger.info("Fetching instances from domain='%s' ...", domain)
 612                 federation.fetch_instances(domain, None, None, inspect.currentframe().f_code.co_name)
 613
 614                 logger.debug("Invoking cookies.clear(%s) ...", domain)
 615                 cookies.clear(domain)
 616             except network.exceptions as exception:
 617                 logger.warning("Exception '%s' during fetching instances (fetch_fba_rss) from domain='%s'", type(exception), domain)
 618                 instances.set_last_error(domain, exception)
 619
 620     logger.debug("Success! - EXIT!")
 621     return 0
 622
 623 def fetch_fbabot_atom(args: argparse.Namespace) -> int:
 624     logger.debug("args[]='%s' - CALLED!", type(args))
 625     feed = "https://ryona.agency/users/fba/feed.atom"
 626
 627     domains = list()
 628
 629     logger.info("Fetching ATOM feed='%s' from FBA bot account ...", feed)
 630     response = utils.fetch_url(feed, network.web_headers, (config.get("connection_timeout"), config.get("read_timeout")))
 631
 632     logger.debug("response.ok='%s',response.status_code=%d,response.text()=%d", response.ok, response.status_code, len(response.text))
 633     if response.ok and response.status_code < 300 and len(response.text) > 0:
 634         logger.debug("Parsing ATOM feed (%d Bytes) ...", len(response.text))
 635         atom = atoma.parse_atom_bytes(response.content)
 636
 637         logger.debug("atom[]='%s'", type(atom))
 638         for entry in atom.entries:
 639             logger.debug("entry[]='%s'", type(entry))
 640             doc = bs4.BeautifulSoup(entry.content.value, "html.parser")
 641             logger.debug("doc[]='%s'", type(doc))
 642             for element in doc.findAll("a"):
 643                 for href in element["href"].split(","):
 644                     logger.debug("href[%s]='%s", type(href), href)
 645                     domain = tidyup.domain(href)
 646
 647                     logger.debug("domain='%s'", domain)
 648                     if not utils.is_domain_wanted(domain):
 649                         logger.debug("domain='%s' is not wanted - SKIPPED!", domain)
 650                         continue
 651                     elif domain in domains:
 652                         logger.debug("domain='%s' is already added - SKIPPED!", domain)
 653                         continue
 654                     elif instances.is_registered(domain):
 655                         logger.debug("domain='%s' is already registered - SKIPPED!", domain)
 656                         continue
 657
 658                     logger.debug("Adding domain='%s',domains()=%d", domain, len(domains))
 659                     domains.append(domain)
 660
 661     logger.debug("domains(%d)='%s", len(domains), domains)
 662     if len(domains) > 0:
 663         locking.acquire()
 664
 665         logger.info("Adding %d new instances ...", len(domains))
 666         for domain in domains:
 667             try:
 668                 logger.info("Fetching instances from domain='%s' ...", domain)
 669                 federation.fetch_instances(domain, None, None, inspect.currentframe().f_code.co_name)
 670
 671                 logger.debug("Invoking cookies.clear(%s) ...", domain)
 672                 cookies.clear(domain)
 673             except network.exceptions as exception:
 674                 logger.warning("Exception '%s' during fetching instances (fetch_fbabot_atom) from domain='%s'", type(exception), domain)
 675                 instances.set_last_error(domain, exception)
 676
 677     logger.debug("Success! - EXIT!")
 678     return 0
 679
 680 def fetch_instances(args: argparse.Namespace) -> int:
 681     logger.debug("args[]='%s' - CALLED!", type(args))
 682     locking.acquire()
 683
 684     # Initial fetch
 685     try:
 686         logger.info("Fetching instances from args.domain='%s' ...", args.domain)
 687         federation.fetch_instances(args.domain, None, None, inspect.currentframe().f_code.co_name)
 688
 689         logger.debug("Invoking cookies.clear(%s) ...", args.domain)
 690         cookies.clear(args.domain)
 691     except network.exceptions as exception:
 692         logger.warning("Exception '%s' during fetching instances (fetch_instances) from args.domain='%s'", type(exception), args.domain)
 693         instances.set_last_error(args.domain, exception)
 694         return 100
 695
 696     if args.single:
 697         logger.debug("Not fetching more instances - EXIT!")
 698         return 0
 699
 700     # Loop through some instances
 701     database.cursor.execute(
 702         "SELECT domain, origin, software, nodeinfo_url FROM instances WHERE software IN ('pleroma', 'mastodon', 'friendica', 'misskey', 'lemmy', 'peertube') AND (last_instance_fetch IS NULL OR last_instance_fetch < ?) ORDER BY rowid DESC", [time.time() - config.get("recheck_instance")]
 703     )
 704
 705     rows = database.cursor.fetchall()
 706     logger.info("Checking %d entries ...", len(rows))
 707     for row in rows:
 708         logger.debug("domain='%s'", row[0])
 709         if blacklist.is_blacklisted(row[0]):
 710             logger.warning("Domain is blacklisted: row[0]='%s'", row[0])
 711             continue
 712
 713         try:
 714             logger.info("Fetching instances for domain='%s',software='%s',origin='%s',nodeinfo_url='%s'", row[0], row[2], row[1], row[3])
 715             federation.fetch_instances(row[0], row[1], row[2], inspect.currentframe().f_code.co_name, row[3])
 716
 717             logger.debug("Invoking cookies.clear(%s) ...", row[0])
 718             cookies.clear(row[0])
 719         except network.exceptions as exception:
 720             logger.warning("Exception '%s' during fetching instances (fetch_instances) from row[0]='%s'", type(exception), row[0])
 721             instances.set_last_error(row[0], exception)
 722
 723     logger.debug("Success - EXIT!")
 724     return 0
 725
 726 def fetch_oliphant(args: argparse.Namespace) -> int:
 727     logger.debug("args[]='%s' - CALLED!", type(args))
 728     locking.acquire()
 729
 730     # Base URL
 731     base_url = "https://codeberg.org/oliphant/blocklists/raw/branch/main/blocklists"
 732
 733     # URLs to fetch
 734     blocklists = (
 735         {
 736             "blocker": "artisan.chat",
 737             "csv_url": "mastodon/artisan.chat.csv",
 738         },{
 739             "blocker": "mastodon.art",
 740             "csv_url": "mastodon/mastodon.art.csv",
 741         },{
 742             "blocker": "pleroma.envs.net",
 743             "csv_url": "mastodon/pleroma.envs.net.csv",
 744         },{
 745             "blocker": "oliphant.social",
 746             "csv_url": "mastodon/_unified_tier3_blocklist.csv",
 747         },{
 748             "blocker": "mastodon.online",
 749             "csv_url": "mastodon/mastodon.online.csv",
 750         },{
 751             "blocker": "mastodon.social",
 752             "csv_url": "mastodon/mastodon.social.csv",
 753         },{
 754             "blocker": "mastodon.social",
 755             "csv_url": "other/missing-tier0-mastodon.social.csv",
 756         },{
 757             "blocker": "rage.love",
 758             "csv_url": "mastodon/rage.love.csv",
 759         },{
 760             "blocker": "sunny.garden",
 761             "csv_url": "mastodon/sunny.garden.csv",
 762         },{
 763             "blocker": "solarpunk.moe",
 764             "csv_url": "mastodon/solarpunk.moe.csv",
 765         },{
 766             "blocker": "toot.wales",
 767             "csv_url": "mastodon/toot.wales.csv",
 768         },{
 769             "blocker": "union.place",
 770             "csv_url": "mastodon/union.place.csv",
 771         }
 772     )
 773
 774     domains = list()
 775
 776     logger.debug("Downloading %d files ...", len(blocklists))
 777     for block in blocklists:
 778         # Is domain given and not equal blocker?
 779         if isinstance(args.domain, str) and args.domain != block["blocker"]:
 780             logger.debug("Skipping blocker='%s', not matching args.domain='%s'", block['blocker'], args.domain)
 781             continue
 782         elif args.domain in domains:
 783             logger.debug("args.domain='%s' already handled - SKIPPED!", args.domain)
 784             continue
 785
 786         # Fetch this URL
 787         logger.info("Fetching csv_url='%s' for blocker='%s' ...", block['csv_url'], block['blocker'])
 788         response = utils.fetch_url(f"{base_url}/{block['csv_url']}", network.web_headers, (config.get("connection_timeout"), config.get("read_timeout")))
 789
 790         logger.debug("response.ok='%s',response.status_code=%d,response.text()=%d", response.ok, response.status_code, len(response.text))
 791         if response.ok and response.content != "":
 792             logger.debug("Fetched %d Bytes, parsing CSV ...", len(response.content))
 793             reader = csv.DictReader(response.content.decode('utf-8').splitlines(), dialect="unix")
 794
 795             logger.debug("reader[]='%s'", type(reader))
 796             for row in reader:
 797                 logger.debug("row[%s]='%s'", type(row), row)
 798                 domain = None
 799                 if "#domain" in row:
 800                     domain = row["#domain"]
 801                 elif "domain" in row:
 802                     domain = row["domain"]
 803                 else:
 804                     logger.debug("row='%s' does not contain domain column", row)
 805                     continue
 806
 807                 logger.debug("domain='%s'", domain)
 808                 if not utils.is_domain_wanted(domain):
 809                     logger.debug("domain='%s' is not wanted - SKIPPED!", domain)
 810                     continue
 811
 812                 logger.debug("Marking domain='%s' as handled", domain)
 813                 domains.append(domain)
 814
 815                 logger.debug("Processing domain='%s' ...", domain)
 816                 processed = utils.process_domain(domain, block["blocker"], inspect.currentframe().f_code.co_name)
 817
 818                 logger.debug("processed='%s'", processed)
 819
 820     logger.debug("Success! - EXIT!")
 821     return 0
 822
 823 def fetch_txt(args: argparse.Namespace) -> int:
 824     logger.debug("args[]='%s' - CALLED!", type(args))
 825     locking.acquire()
 826
 827     # Static URLs
 828     urls = (
 829         "https://seirdy.one/pb/bsl.txt",
 830     )
 831
 832     logger.info("Checking %d text file(s) ...", len(urls))
 833     for url in urls:
 834         logger.debug("Fetching url='%s' ...", url)
 835         response = utils.fetch_url(url, network.web_headers, (config.get("connection_timeout"), config.get("read_timeout")))
 836
 837         logger.debug("response.ok='%s',response.status_code=%d,response.text()=%d", response.ok, response.status_code, len(response.text))
 838         if response.ok and response.status_code < 300 and response.text != "":
 839             logger.debug("Returned %d Bytes for processing", len(response.text.strip()))
 840             domains = response.text.split("\n")
 841
 842             logger.info("Processing %d domains ...", len(domains))
 843             for domain in domains:
 844                 logger.debug("domain='%s'", domain)
 845                 if domain == "":
 846                     logger.debug("domain is empty - SKIPPED!")
 847                     continue
 848                 elif not utils.is_domain_wanted(domain):
 849                     logger.debug("domain='%s' is not wanted - SKIPPED!", domain)
 850                     continue
 851
 852                 logger.debug("domain='%s'", domain)
 853                 processed = utils.process_domain(domain, 'seirdy.one', inspect.currentframe().f_code.co_name)
 854
 855                 logger.debug("processed='%s'", processed)
 856                 if not processed:
 857                     logger.debug("domain='%s' was not generically processed - SKIPPED!", domain)
 858                     continue
 859
 860     logger.debug("Success! - EXIT!")
 861     return 0
 862
 863 def fetch_fedipact(args: argparse.Namespace) -> int:
 864     logger.debug("args[]='%s' - CALLED!", type(args))
 865     locking.acquire()
 866
 867     response = utils.fetch_url("https://fedipact.online", network.web_headers, (config.get("connection_timeout"), config.get("read_timeout")))
 868
 869     logger.debug("response.ok='%s',response.status_code=%d,response.text()=%d", response.ok, response.status_code, len(response.text))
 870     if response.ok and response.status_code < 300 and response.text != "":
 871         logger.debug("Parsing %d Bytes ...", len(response.text))
 872
 873         doc = bs4.BeautifulSoup(response.text, "html.parser")
 874         logger.debug("doc[]='%s'", type(doc))
 875
 876         rows = doc.findAll("li")
 877         logger.info("Checking %d row(s) ...", len(rows))
 878         for row in rows:
 879             logger.debug("row[]='%s'", type(row))
 880             domain = tidyup.domain(row.contents[0])
 881
 882             logger.debug("domain='%s'", domain)
 883             if domain == "":
 884                 logger.debug("domain is empty - SKIPPED!")
 885                 continue
 886             elif not utils.is_domain_wanted(domain):
 887                 logger.debug("domain='%s' is not wanted - SKIPPED!", domain)
 888                 continue
 889             elif instances.is_registered(domain):
 890                 logger.debug("domain='%s' is already registered - SKIPPED!", domain)
 891                 continue
 892
 893             logger.info("Fetching domain='%s' ...", domain)
 894             federation.fetch_instances(domain, None, None, inspect.currentframe().f_code.co_name)
 895
 896     logger.debug("Success! - EXIT!")
 897     return 0