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 from urllib.parse import urlparse
  24
  25 import argparse
  26 import atoma
  27 import bs4
  28 import markdown
  29 import reqto
  30 import validators
  31
  32 from fba import database
  33 from fba import utils
  34
  35 from fba.helpers import blacklist
  36 from fba.helpers import blocklists
  37 from fba.helpers import config
  38 from fba.helpers import cookies
  39 from fba.helpers import dicts as dict_helper
  40 from fba.helpers import domain as domain_helper
  41 from fba.helpers import locking
  42 from fba.helpers import processing
  43 from fba.helpers import software as software_helper
  44 from fba.helpers import tidyup
  45
  46 from fba.http import csrf
  47 from fba.http import federation
  48 from fba.http import network
  49
  50 from fba.models import blocks
  51 from fba.models import instances
  52 from fba.models import sources
  53
  54 from fba.networks import friendica
  55 from fba.networks import lemmy
  56 from fba.networks import mastodon
  57 from fba.networks import misskey
  58 from fba.networks import pleroma
  59
  60 logging.basicConfig(level=logging.INFO)
  61 logger = logging.getLogger(__name__)
  62 #logger.setLevel(logging.DEBUG)
  63
  64 def check_instance(args: argparse.Namespace) -> int:
  65     logger.debug("args.domain='%s' - CALLED!", args.domain)
  66
  67     status = 0
  68     if not validators.domain(args.domain):
  69         logger.warning("args.domain='%s' is not valid", args.domain)
  70         status = 100
  71     elif blacklist.is_blacklisted(args.domain):
  72         logger.warning("args.domain='%s' is blacklisted", args.domain)
  73         status = 101
  74     elif instances.is_registered(args.domain):
  75         logger.warning("args.domain='%s' is already registered", args.domain)
  76         status = 102
  77     else:
  78         logger.info("args.domain='%s' is not known", args.domain)
  79
  80     logger.debug("status=%d - EXIT!", status)
  81     return status
  82
  83 def check_nodeinfo(args: argparse.Namespace) -> int:
  84     logger.debug("args[]='%s' - CALLED!", type(args))
  85
  86     # Fetch rows
  87     database.cursor.execute("SELECT domain, software, nodeinfo_url FROM instances WHERE nodeinfo_url IS NOT NULL ORDER BY domain ASC")
  88
  89     cnt = 0
  90     for row in database.cursor.fetchall():
  91         logger.debug("Checking row[domain]='%s',row[software]='%s',row[nodeinfo_url]='%s' ...", row["domain"], row["software"], row["nodeinfo_url"])
  92         punycode = row["domain"].encode("idna").decode("utf-8")
  93
  94         if row["nodeinfo_url"].startswith("/"):
  95             logger.debug("row[nodeinfo_url]='%s' is a relative URL and always matches", row["nodeinfo_url"])
  96             continue
  97         elif row["nodeinfo_url"].find(punycode) == -1 and row["nodeinfo_url"].find(row["domain"]) == -1:
  98             logger.warning("punycode='%s' is not found in row[nodeinfo_url]='%s',row[software]='%s'", punycode, row["nodeinfo_url"], row["software"])
  99             cnt = cnt + 1
 100
 101     logger.info("Found %d row(s)", cnt)
 102
 103     logger.debug("EXIT!")
 104     return 0
 105
 106 def fetch_pixelfed_api(args: argparse.Namespace) -> int:
 107     logger.debug("args[]='%s' - CALLED!", type(args))
 108
 109     # No CSRF by default, you don't have to add network.source_headers by yourself here
 110     headers = tuple()
 111     source_domain = "pixelfed.org"
 112
 113     if sources.is_recent(source_domain):
 114         logger.info("API from source_domain='%s' has recently being accessed - EXIT!", source_domain)
 115         return 1
 116     else:
 117         logger.debug("source_domain='%s' has not been recently used, marking ...", source_domain)
 118         sources.update(source_domain)
 119
 120     try:
 121         logger.debug("Checking CSRF from source_domain='%s' ...", source_domain)
 122         headers = csrf.determine(source_domain, dict())
 123     except network.exceptions as exception:
 124         logger.warning("Exception '%s' during checking CSRF (fetch_peers,%s) - EXIT!", type(exception), __name__)
 125         return list()
 126
 127     try:
 128         logger.info("Fetching JSON from pixelfed.org API, headers()=%d ...", len(headers))
 129         fetched = network.get_json_api(
 130             source_domain,
 131             "/api/v1/servers/all.json?scope=All&country=all&language=all",
 132             headers,
 133             (config.get("connection_timeout"), config.get("read_timeout"))
 134         )
 135
 136         logger.debug("JSON API returned %d elements", len(fetched))
 137         if "error_message" in fetched:
 138             logger.warning("API returned error_message='%s' - EXIT!", fetched["error_message"])
 139             return 101
 140         elif "data" not in fetched["json"]:
 141             logger.warning("API did not return JSON with 'data' element - EXIT!")
 142             return 102
 143
 144         rows = fetched["json"]["data"]
 145         logger.info("Checking %d fetched rows ...", len(rows))
 146         for row in rows:
 147             logger.debug("row[]='%s'", type(row))
 148             if "domain" not in row:
 149                 logger.warning("row='%s' does not contain element 'domain' - SKIPPED!", row)
 150                 continue
 151             elif row["domain"] is None or row["domain"] == "":
 152                 logger.debug("row[domain]='%s' is empty - SKIPPED!", row["domain"])
 153                 continue
 154
 155             logger.debug("row[domain]='%s' - BEFORE!", row["domain"])
 156             domain = row["domain"].encode("idna").decode("utf-8")
 157             logger.debug("domain='%s' - AFTER!", domain)
 158
 159             if not domain_helper.is_wanted(domain):
 160                 logger.debug("domain='%s' is not wanted - SKIPPED!", domain)
 161                 continue
 162             elif instances.is_registered(domain):
 163                 logger.debug("domain='%s' is already registered - SKIPPED!", domain)
 164                 continue
 165             elif instances.is_recent(domain):
 166                 logger.debug("domain='%s' has been recently crawled - SKIPPED!", domain)
 167                 continue
 168
 169             logger.debug("Fetching instances from domain='%s' ...", domain)
 170             federation.fetch_instances(domain, None, None, inspect.currentframe().f_code.co_name)
 171
 172     except network.exceptions as exception:
 173         logger.warning("Cannot fetch graphql,exception[%s]:'%s' - EXIT!", type(exception), str(exception))
 174         return 103
 175
 176     logger.debug("Success! - EXIT!")
 177     return 0
 178
 179 def fetch_bkali(args: argparse.Namespace) -> int:
 180     logger.debug("args[]='%s' - CALLED!", type(args))
 181
 182     logger.debug("Invoking locking.acquire() ...")
 183     locking.acquire()
 184
 185     source_domain = "gql.api.bka.li"
 186     if sources.is_recent(source_domain):
 187         logger.info("API from source_domain='%s' has recently being accessed - EXIT!", source_domain)
 188         return 1
 189     else:
 190         logger.debug("source_domain='%s' has not been recently used, marking ...", source_domain)
 191         sources.update(source_domain)
 192
 193     domains = list()
 194     try:
 195         logger.info("Fetching domainlist from source_domain='%s' ...", source_domain)
 196         fetched = network.post_json_api(
 197             source_domain,
 198             "/v1/graphql",
 199             json.dumps({
 200                 "query": "query domainlist {nodeinfo(order_by: {domain: asc}) {domain}}"
 201             })
 202         )
 203
 204         logger.debug("fetched[]='%s'", type(fetched))
 205         if "error_message" in fetched:
 206             logger.warning("post_json_api() for 'gql.sources.bka.li' returned error message='%s", fetched["error_message"])
 207             return 100
 208         elif isinstance(fetched["json"], dict) and "error" in fetched["json"] and "message" in fetched["json"]["error"]:
 209             logger.warning("post_json_api() returned error: '%s", fetched["error"]["message"])
 210             return 101
 211
 212         rows = fetched["json"]
 213
 214         logger.debug("rows(%d)[]='%s'", len(rows), type(rows))
 215         if len(rows) == 0:
 216             raise Exception("WARNING: Returned no records")
 217         elif "data" not in rows:
 218             raise Exception(f"WARNING: rows()={len(rows)} does not contain key 'data'")
 219         elif "nodeinfo" not in rows["data"]:
 220             raise Exception(f"WARNING: rows()={len(rows['data'])} does not contain key 'nodeinfo'")
 221
 222         for entry in rows["data"]["nodeinfo"]:
 223             logger.debug("entry[%s]='%s'", type(entry), entry)
 224             if "domain" not in entry:
 225                 logger.warning("entry()=%d does not contain 'domain' - SKIPPED!", len(entry))
 226                 continue
 227             elif entry["domain"] is None or entry["domain"] == "":
 228                 logger.debug("entry[domain]='%s' is empty - SKIPPED!", entry["domain"])
 229                 continue
 230             elif not domain_helper.is_wanted(entry["domain"]):
 231                 logger.debug("entry[domain]='%s' is not wanted - SKIPPED!", entry["domain"])
 232                 continue
 233             elif instances.is_registered(entry["domain"]):
 234                 logger.debug("entry[domain]='%s' is already registered - SKIPPED!", entry["domain"])
 235                 continue
 236             elif instances.is_recent(entry["domain"]):
 237                 logger.debug("entry[domain]='%s' has been recently crawled - SKIPPED!", entry["domain"])
 238                 continue
 239
 240             logger.debug("Adding domain='%s' ...", entry["domain"])
 241             domains.append(entry["domain"])
 242
 243     except network.exceptions as exception:
 244         logger.warning("Cannot fetch graphql,exception[%s]:'%s' - EXIT!", type(exception), str(exception))
 245         return 102
 246
 247     logger.debug("domains()=%d", len(domains))
 248     if len(domains) > 0:
 249         logger.info("Adding %d new instances ...", len(domains))
 250         for domain in domains:
 251             logger.debug("domain='%s' - BEFORE!", domain)
 252             domain = domain.encode("idna").decode("utf-8")
 253             logger.debug("domain='%s' - AFTER!", domain)
 254
 255             try:
 256                 logger.info("Fetching instances from domain='%s' ...", domain)
 257                 federation.fetch_instances(domain, 'tak.teleyal.blog', None, inspect.currentframe().f_code.co_name)
 258             except network.exceptions as exception:
 259                 logger.warning("Exception '%s' during fetching instances (fetch_bkali) from domain='%s'", type(exception), domain)
 260                 instances.set_last_error(domain, exception)
 261                 return 100
 262
 263     logger.debug("Success - EXIT!")
 264     return 0
 265
 266 def fetch_blocks(args: argparse.Namespace) -> int:
 267     logger.debug("args[]='%s' - CALLED!", type(args))
 268     if args.domain is not None and args.domain != "":
 269         logger.debug("args.domain='%s' - checking ...", args.domain)
 270         if not validators.domain(args.domain):
 271             logger.warning("args.domain='%s' is not valid.", args.domain)
 272             return 100
 273         elif blacklist.is_blacklisted(args.domain):
 274             logger.warning("args.domain='%s' is blacklisted, won't check it!", args.domain)
 275             return 101
 276         elif not instances.is_registered(args.domain):
 277             logger.warning("args.domain='%s' is not registered, please run ./utils.py fetch_instances '%s' first.", args.domain, args.domain)
 278             return 102
 279
 280     logger.debug("Invoking locking.acquire() ...")
 281     locking.acquire()
 282
 283     if args.domain is not None and args.domain != "":
 284         # Re-check single domain
 285         logger.debug("Querying database for args.domain='%s' ...", args.domain)
 286         database.cursor.execute(
 287             "SELECT domain, software, origin, nodeinfo_url FROM instances WHERE domain = ? LIMIT 1", [args.domain]
 288         )
 289     elif args.software is not None and args.software != "":
 290         # Re-check single software
 291         logger.debug("Querying database for args.software='%s' ...", args.software)
 292         database.cursor.execute(
 293             "SELECT domain, software, origin, nodeinfo_url FROM instances WHERE software = ? AND nodeinfo_url IS NOT NULL ORDER BY total_blocks DESC, last_response_time ASC, last_updated ASC", [args.software]
 294         )
 295     elif args.only_none:
 296         # Check only entries with total_blocked=None
 297         database.cursor.execute(
 298             "SELECT domain, software, origin, nodeinfo_url FROM instances WHERE software IN ('pleroma', 'mastodon', 'lemmy', 'friendica', 'misskey') AND total_blocks IS NULL ORDER BY total_blocks DESC, last_response_time ASC, last_updated ASC"
 299         )
 300     else:
 301         # Re-check after "timeout" (aka. minimum interval)
 302         database.cursor.execute(
 303             "SELECT domain, software, origin, nodeinfo_url FROM instances WHERE software IN ('pleroma', 'mastodon', 'lemmy', 'friendica', 'misskey') AND nodeinfo_url IS NOT NULL ORDER BY total_blocks DESC, last_response_time ASC, last_updated ASC"
 304         )
 305
 306     rows = database.cursor.fetchall()
 307     logger.info("Checking %d entries ...", len(rows))
 308     for blocker, software, origin, nodeinfo_url in rows:
 309         logger.debug("blocker='%s',software='%s',origin='%s',nodeinfo_url='%s'", blocker, software, origin, nodeinfo_url)
 310
 311         if not domain_helper.is_wanted(blocker):
 312             logger.warning("blocker='%s' is not wanted - SKIPPED!", blocker)
 313             continue
 314         elif not args.force and instances.is_recent(blocker, "last_blocked"):
 315             logger.debug("blocker='%s' has been recently accessed - SKIPPED!", blocker)
 316             continue
 317
 318         logger.debug("Setting last_blocked,has_obfuscation=false for blocker='%s' ...", blocker)
 319         instances.set_last_blocked(blocker)
 320         instances.set_has_obfuscation(blocker, False)
 321
 322         # c.s isn't part of oliphant's "hidden" blocklists
 323         if blocker == "chaos.social" or software_helper.is_relay(software) or blocklists.has(blocker):
 324             logger.debug("Skipping blocker='%s', run ./fba.py fetch_cs, fetch_oliphant, fetch_csv instead!", blocker)
 325             continue
 326
 327         logger.debug("Invoking federation.fetch_blocks(%s) ...", blocker)
 328         blocking = federation.fetch_blocks(blocker)
 329
 330         logger.debug("blocker='%s',software='%s',blocking()=%d", blocker, software, len(blocking))
 331         if len(blocking) == 0:
 332             logger.debug("blocker='%s',software='%s' - fetching blocklist ...", blocker, software)
 333             if software == "pleroma":
 334                 blocking = pleroma.fetch_blocks(blocker)
 335                 logger.debug("blocker='%s' returned %d entries,software='%s'", blocker, len(blocking), software)
 336             elif software == "mastodon":
 337                 blocking = mastodon.fetch_blocks(blocker)
 338                 logger.debug("blocker='%s' returned %d entries,software='%s'", blocker, len(blocking), software)
 339             elif software == "lemmy":
 340                 blocking = lemmy.fetch_blocks(blocker)
 341                 logger.debug("blocker='%s' returned %d entries,software='%s'", blocker, len(blocking), software)
 342             elif software == "friendica":
 343                 blocking = friendica.fetch_blocks(blocker)
 344                 logger.debug("blocker='%s' returned %d entries,software='%s'", blocker, len(blocking), software)
 345             elif software == "misskey":
 346                 blocking = misskey.fetch_blocks(blocker)
 347                 logger.debug("blocker='%s' returned %d entries,software='%s'", blocker, len(blocking), software)
 348             else:
 349                 logger.warning("Unknown software: blocker='%s',software='%s'", blocker, software)
 350
 351         logger.debug("Invoking instances.set_total_blocks(%s, %d) ...", blocker, len(blocking))
 352         instances.set_total_blocks(blocker, blocking)
 353
 354         blockdict = list()
 355         deobfuscated = obfuscated = 0
 356
 357         logger.info("Checking %d entries from blocker='%s',software='%s' ...", len(blocking), blocker, software)
 358         for block in blocking:
 359             logger.debug("blocked='%s',block_level='%s',reason='%s'", block["blocked"], block["block_level"], block["reason"])
 360
 361             if block["block_level"] == "":
 362                 logger.warning("block_level is empty, blocker='%s',blocked='%s'", block["blocker"], block["blocked"])
 363                 continue
 364
 365             logger.debug("blocked='%s',reason='%s' - BEFORE!", block["blocked"], block["reason"])
 366             block["blocked"] = tidyup.domain(block["blocked"])
 367             block["reason"]  = tidyup.reason(block["reason"]) if block["reason"] is not None and block["reason"] != "" else None
 368             logger.debug("blocked='%s',reason='%s' - AFTER!", block["blocked"], block["reason"])
 369
 370             if block["blocked"] is None or block["blocked"] == "":
 371                 logger.warning("block[blocked]='%s' is empty, blocker='%s'", block["blocked"], blocker)
 372                 continue
 373             elif block["blocked"].endswith(".onion"):
 374                 logger.debug("blocked='%s' is a TOR .onion domain - SKIPPED", block["blocked"])
 375                 continue
 376             elif block["blocked"].endswith(".i2p") and config.get("allow_i2p_domain") == "true":
 377                 logger.debug("blocked='%s' is an I2P .onion domain - SKIPPED", block["blocked"])
 378                 continue
 379             elif block["blocked"].endswith(".arpa"):
 380                 logger.debug("blocked='%s' is a reverse IP address - SKIPPED", block["blocked"])
 381                 continue
 382             elif block["blocked"].endswith(".tld"):
 383                 logger.debug("blocked='%s' is a fake domain - SKIPPED", block["blocked"])
 384                 continue
 385             elif block["blocked"].find("*") >= 0:
 386                 logger.debug("blocker='%s' uses obfuscated domains", blocker)
 387                 instances.set_has_obfuscation(blocker, True)
 388                 obfuscated = obfuscated + 1
 389
 390                 # Some friendica servers also obscure domains without hash
 391                 row = instances.deobfuscate("*", block["blocked"], block["digest"] if "digest" in block else None)
 392
 393                 logger.debug("row[]='%s'", type(row))
 394                 if row is None:
 395                     logger.warning("Cannot deobfuscate blocked='%s',blocker='%s',software='%s' - SKIPPED!", block["blocked"], blocker, software)
 396                     continue
 397
 398                 deobfuscated = deobfuscated + 1
 399                 block["blocked"] = row["domain"]
 400                 origin           = row["origin"]
 401                 nodeinfo_url     = row["nodeinfo_url"]
 402             elif block["blocked"].find("?") >= 0:
 403                 logger.debug("blocker='%s' uses obfuscated domains", blocker)
 404                 instances.set_has_obfuscation(blocker, True)
 405                 obfuscated = obfuscated + 1
 406
 407                 # Some obscure them with question marks, not sure if that's dependent on version or not
 408                 row = instances.deobfuscate("?", block["blocked"], block["digest"] if "digest" in block else None)
 409
 410                 logger.debug("row[]='%s'", type(row))
 411                 if row is None:
 412                     logger.warning("Cannot deobfuscate blocked='%s',blocker='%s',software='%s' - SKIPPED!", block["blocked"], blocker, software)
 413                     continue
 414
 415                 deobfuscated = deobfuscated + 1
 416                 block["blocked"] = row["domain"]
 417                 origin           = row["origin"]
 418                 nodeinfo_url     = row["nodeinfo_url"]
 419
 420             logger.debug("Looking up instance by domain, blocked='%s'", block["blocked"])
 421             if block["blocked"] is None or block["blocked"] == "":
 422                 logger.debug("block[blocked]='%s' is empty - SKIPPED!", block["blocked"])
 423                 continue
 424
 425             logger.debug("block[blocked]='%s' - BEFORE!", block["blocked"])
 426             block["blocked"] = block["blocked"].lstrip(".").encode("idna").decode("utf-8")
 427             logger.debug("block[blocked]='%s' - AFTER!", block["blocked"])
 428
 429             if not domain_helper.is_wanted(block["blocked"]):
 430                 logger.debug("blocked='%s' is not wanted - SKIPPED!", block["blocked"])
 431                 continue
 432             elif block["block_level"] in ["accept", "accepted"]:
 433                 logger.debug("blocked='%s' is accepted, not wanted here - SKIPPED!", block["blocked"])
 434                 continue
 435             elif not instances.is_registered(block["blocked"]):
 436                 logger.debug("Hash wasn't found, adding: blocked='%s',blocker='%s'", block["blocked"], blocker)
 437                 federation.fetch_instances(block["blocked"], blocker, None, inspect.currentframe().f_code.co_name)
 438
 439             block["block_level"] = blocks.alias_block_level(block["block_level"])
 440
 441             if processing.block(blocker, block["blocked"], block["reason"], block["block_level"]) and block["block_level"] == "reject" and config.get("bot_enabled"):
 442                 logger.debug("Appending blocked='%s',reason='%s' for blocker='%s' ...", block["blocked"], block["block_level"], blocker)
 443                 blockdict.append({
 444                     "blocked": block["blocked"],
 445                     "reason" : block["reason"],
 446                 })
 447
 448             logger.debug("Invoking cookies.clear(%s) ...", block["blocked"])
 449             cookies.clear(block["blocked"])
 450
 451         logger.info("blocker='%s' has %d obfuscated domain(s) and %d of them could be deobfuscated.", blocker, obfuscated, deobfuscated)
 452         instances.set_obfuscated_blocks(blocker, obfuscated)
 453
 454         logger.debug("Flushing updates for blocker='%s' ...", blocker)
 455         instances.update(blocker)
 456
 457         logger.debug("Invoking commit() ...")
 458         database.connection.commit()
 459
 460         logger.debug("Invoking cookies.clear(%s) ...", blocker)
 461         cookies.clear(blocker)
 462
 463         logger.debug("config.get(bot_enabled)='%s',blockdict()=%d'", config.get("bot_enabled"), len(blockdict))
 464         if config.get("bot_enabled") and len(blockdict) > 0:
 465             logger.info("Sending bot POST for blocker='%s',blockdict()=%d ...", blocker, len(blockdict))
 466             network.send_bot_post(blocker, blockdict)
 467
 468     logger.debug("Success! - EXIT!")
 469     return 0
 470
 471 def fetch_observer(args: argparse.Namespace) -> int:
 472     logger.debug("args[]='%s' - CALLED!", type(args))
 473
 474     logger.debug("Invoking locking.acquire() ...")
 475     locking.acquire()
 476
 477     source_domain = "fediverse.observer"
 478     if sources.is_recent(source_domain):
 479         logger.info("API from source_domain='%s' has recently being accessed - EXIT!", source_domain)
 480         return 1
 481     else:
 482         logger.debug("source_domain='%s' has not been recently used, marking ...", source_domain)
 483         sources.update(source_domain)
 484
 485     types = list()
 486     if args.software is None:
 487         logger.info("Fetching software list ...")
 488         raw = utils.fetch_url(
 489             f"https://{source_domain}",
 490             network.web_headers,
 491             (config.get("connection_timeout"), config.get("read_timeout"))
 492         ).text
 493         logger.debug("raw[%s]()=%d", type(raw), len(raw))
 494
 495         doc = bs4.BeautifulSoup(raw, features="html.parser")
 496         logger.debug("doc[]='%s'", type(doc))
 497
 498         navbar = doc.find("div", {"aria-labelledby": "navbarDropdownMenuSoftwares"})
 499         logger.debug("navbar[]='%s'", type(navbar))
 500         if navbar is None:
 501             logger.warning("Cannot find navigation bar, cannot continue!")
 502             return 1
 503
 504         items = navbar.findAll("a", {"class": "dropdown-item"})
 505         logger.debug("items[]='%s'", type(items))
 506
 507         logger.info("Checking %d menu items ...", len(items))
 508         for item in items:
 509             logger.debug("item[%s]='%s'", type(item), item)
 510             if item.text.lower() == "all":
 511                 logger.debug("Skipping 'All' menu entry ...")
 512                 continue
 513
 514             logger.debug("Appending item.text='%s' ...", item.text)
 515             types.append(tidyup.domain(item.text))
 516     else:
 517         logger.info("Adding args.software='%s' as type ...", args.software)
 518         types.append(args.software)
 519
 520     logger.info("Fetching %d different table data ...", len(types))
 521     for software in types:
 522         logger.debug("software='%s'", software)
 523
 524         if args.software is not None and args.software != software:
 525             logger.debug("args.software='%s' does not match software='%s' - SKIPPED!", args.software, software)
 526             continue
 527
 528         doc = None
 529         try:
 530             logger.debug("Fetching table data for software='%s' ...", software)
 531             raw = utils.fetch_url(
 532                 f"https://{source_domain}/app/views/tabledata.php?software={software}",
 533                 network.web_headers,
 534                 (config.get("connection_timeout"), config.get("read_timeout"))
 535             ).text
 536             logger.debug("raw[%s]()=%d", type(raw), len(raw))
 537
 538             doc = bs4.BeautifulSoup(raw, features="html.parser")
 539             logger.debug("doc[]='%s'", type(doc))
 540         except network.exceptions as exception:
 541             logger.warning("Cannot fetch software='%s' from source_domain='%s': '%s'", software, source_domain, type(exception))
 542             continue
 543
 544         items = doc.findAll("a", {"class": "url"})
 545         logger.info("Checking %d items,software='%s' ...", len(items), software)
 546         for item in items:
 547             logger.debug("item[]='%s'", type(item))
 548             domain = item.decode_contents()
 549             logger.debug("domain[%s]='%s'", type(domain), domain)
 550             domain = tidyup.domain(domain) if domain not in [None, ""] else None
 551             logger.debug("domain='%s' - AFTER!", domain)
 552
 553             if domain is None or domain == "":
 554                 logger.debug("domain='%s' is empty after tidyup.domain() - SKIPPED!", domain)
 555                 continue
 556
 557             logger.debug("domain='%s' - BEFORE!", domain)
 558             domain = domain.encode("idna").decode("utf-8")
 559             logger.debug("domain='%s' - AFTER!", domain)
 560
 561             if not domain_helper.is_wanted(domain):
 562                 logger.debug("domain='%s' is not wanted - SKIPPED!", domain)
 563                 continue
 564             elif instances.is_registered(domain):
 565                 logger.debug("domain='%s' is already registered - SKIPPED!", domain)
 566                 continue
 567
 568             logger.info("Fetching instances for domain='%s'", domain)
 569             federation.fetch_instances(domain, None, None, inspect.currentframe().f_code.co_name)
 570
 571     logger.debug("Success! - EXIT!")
 572     return 0
 573
 574 def fetch_todon_wiki(args: argparse.Namespace) -> int:
 575     logger.debug("args[]='%s' - CALLED!", type(args))
 576
 577     logger.debug("Invoking locking.acquire() ...")
 578     locking.acquire()
 579
 580     source_domain = "wiki.todon.eu"
 581     if sources.is_recent(source_domain):
 582         logger.info("API from source_domain='%s' has recently being accessed - EXIT!", source_domain)
 583         return 1
 584     else:
 585         logger.debug("source_domain='%s' has not been recently used, marking ...", source_domain)
 586         sources.update(source_domain)
 587
 588     blocklist = {
 589         "silenced": list(),
 590         "reject": list(),
 591     }
 592
 593     logger.debug("Fetching domainblocks from source_domain='%s'", source_domain)
 594     raw = utils.fetch_url(
 595         f"https://{source_domain}/todon/domainblocks",
 596         network.web_headers,
 597         (config.get("connection_timeout"), config.get("read_timeout"))
 598     ).text
 599     logger.debug("raw()=%d,raw[]='%s'", len(raw), type(raw))
 600
 601     doc = bs4.BeautifulSoup(raw, "html.parser")
 602     logger.debug("doc[]='%s'", type(doc))
 603
 604     silenced = doc.find("h3", {"id": "silencedlimited_servers"}).find_next("ul").findAll("li")
 605     logger.info("Checking %d silenced/limited entries ...", len(silenced))
 606     blocklist["silenced"] = utils.find_domains(silenced, "div")
 607
 608     suspended = doc.find("h3", {"id": "suspended_servers"}).find_next("ul").findAll("li")
 609     logger.info("Checking %d suspended entries ...", len(suspended))
 610     blocklist["reject"] = utils.find_domains(suspended, "div")
 611
 612     blocking = blocklist["silenced"] + blocklist["reject"]
 613     blocker = "todon.eu"
 614
 615     logger.debug("Invoking instances.set_total_blocks(%s, %d) ...", blocker, len(blocking))
 616     instances.set_last_blocked(blocker)
 617     instances.set_total_blocks(blocker, blocking)
 618
 619     blockdict = list()
 620     for block_level in blocklist:
 621         blockers = blocklist[block_level]
 622
 623         logger.debug("block_level='%s',blockers()=%d'", block_level, len(blockers))
 624         for blocked in blockers:
 625             logger.debug("blocked='%s'", blocked)
 626
 627             if not instances.is_registered(blocked):
 628                 try:
 629                     logger.info("Fetching instances from domain='%s' ...", blocked)
 630                     federation.fetch_instances(blocked, blocker, None, inspect.currentframe().f_code.co_name)
 631                 except network.exceptions as exception:
 632                     logger.warning("Exception '%s' during fetching instances (fetch_cs) from blocked='%s'", type(exception), blocked)
 633                     instances.set_last_error(blocked, exception)
 634
 635             if not domain_helper.is_wanted(blocked):
 636                 logger.warning("blocked='%s' is not wanted - SKIPPED!", blocked)
 637                 continue
 638             elif not domain_helper.is_wanted(blocker):
 639                 logger.warning("blocker='%s' is not wanted - SKIPPED!", blocker)
 640                 continue
 641             elif blocks.is_instance_blocked(blocker, blocked, block_level):
 642                 logger.debug("blocked='%s',block_level='%s' is already blocked - SKIPPED!", blocked, block_level)
 643                 continue
 644
 645             logger.info("Adding new block: blocked='%s',block_level='%s'", blocked, block_level)
 646             if processing.block(blocker, blocked, None, block_level) and block_level == "reject" and config.get("bot_enabled"):
 647                 logger.debug("Appending blocked='%s',reason='%s' for blocker='%s' ...", blocked, block_level, blocker)
 648                 blockdict.append({
 649                     "blocked": blocked,
 650                     "reason" : None,
 651                 })
 652
 653         logger.debug("Invoking commit() ...")
 654         database.connection.commit()
 655
 656         logger.debug("config.get(bot_enabled)='%s',blockdict()=%d", config.get("bot_enabled"), len(blockdict))
 657         if config.get("bot_enabled") and len(blockdict) > 0:
 658             logger.info("Sending bot POST for blocker='%s',blockdict()=%d ...", blocker, len(blockdict))
 659             network.send_bot_post(blocker, blockdict)
 660
 661     logger.debug("Checking if blocker='%s' has pending updates ...", blocker)
 662     if instances.has_pending(blocker):
 663         logger.debug("Flushing updates for blocker='%s' ...", blocker)
 664         instances.update(blocker)
 665
 666     logger.debug("Success! - EXIT!")
 667     return 0
 668
 669 def fetch_cs(args: argparse.Namespace):
 670     logger.debug("args[]='%s' - CALLED!", type(args))
 671
 672     logger.debug("Invoking locking.acquire() ...")
 673     locking.acquire()
 674
 675     extensions = [
 676         "extra",
 677         "abbr",
 678         "attr_list",
 679         "def_list",
 680         "fenced_code",
 681         "footnotes",
 682         "md_in_html",
 683         "admonition",
 684         "codehilite",
 685         "legacy_attrs",
 686         "legacy_em",
 687         "meta",
 688         "nl2br",
 689         "sane_lists",
 690         "smarty",
 691         "toc",
 692         "wikilinks"
 693     ]
 694
 695     blocklist = {
 696         "silenced": list(),
 697         "reject"  : list(),
 698     }
 699
 700     source_domain = "raw.githubusercontent.com"
 701     if sources.is_recent(source_domain):
 702         logger.info("API from source_domain='%s' has recently being accessed - EXIT!", source_domain)
 703         return 1
 704     else:
 705         logger.debug("source_domain='%s' has not been recently used, marking ...", source_domain)
 706         sources.update(source_domain)
 707
 708     logger.info("Fetching federation.md from source_domain='%s' ...", source_domain)
 709     raw = utils.fetch_url(
 710         f"https://{source_domain}/chaossocial/meta/master/federation.md",
 711         network.web_headers,
 712         (config.get("connection_timeout"), config.get("read_timeout"))
 713     ).text
 714     logger.debug("raw()=%d,raw[]='%s'", len(raw), type(raw))
 715
 716     doc = bs4.BeautifulSoup(markdown.markdown(raw, extensions=extensions), features="html.parser")
 717     logger.debug("doc()=%d[]='%s'", len(doc), type(doc))
 718
 719     silenced = doc.find("h2", {"id": "silenced-instances"}).findNext("table").find("tbody")
 720     logger.debug("silenced[%s]()=%d", type(silenced), len(silenced))
 721     blocklist["silenced"] = federation.find_domains(silenced)
 722
 723     blocked = doc.find("h2", {"id": "blocked-instances"}).findNext("table").find("tbody")
 724     logger.debug("blocked[%s]()=%d", type(blocked), len(blocked))
 725     blocklist["reject"] = federation.find_domains(blocked)
 726
 727     blocking = blocklist["silenced"] + blocklist["reject"]
 728     blocker = "chaos.social"
 729
 730     logger.debug("Invoking instances.set_total_blocks(%s, %d) ...", blocker, len(blocking))
 731     instances.set_last_blocked(blocker)
 732     instances.set_total_blocks(blocker, blocking)
 733
 734     logger.debug("blocklist[silenced]()=%d,blocklist[reject]()=%d", len(blocklist["silenced"]), len(blocklist["reject"]))
 735     if len(blocking) > 0:
 736         blockdict = list()
 737         for block_level in blocklist:
 738             logger.info("block_level='%s' has %d row(s)", block_level, len(blocklist[block_level]))
 739
 740             for row in blocklist[block_level]:
 741                 logger.debug("row[%s]='%s'", type(row), row)
 742                 if not "domain" in row:
 743                     logger.warning("row[]='%s' has no element 'domain' - SKIPPED!", type(row))
 744                     continue
 745                 elif not instances.is_registered(row["domain"]):
 746                     try:
 747                         logger.info("Fetching instances from domain='%s' ...", row["domain"])
 748                         federation.fetch_instances(row["domain"], blocker, None, inspect.currentframe().f_code.co_name)
 749                     except network.exceptions as exception:
 750                         logger.warning("Exception '%s' during fetching instances (fetch_cs) from row[domain]='%s'", type(exception), row["domain"])
 751                         instances.set_last_error(row["domain"], exception)
 752
 753                 if processing.block(blocker, row["domain"], row["reason"], block_level) and block_level == "reject" and config.get("bot_enabled"):
 754                     logger.debug("Appending blocked='%s',reason='%s' for blocker='%s' ...", row["domain"], block_level, blocker)
 755                     blockdict.append({
 756                         "blocked": row["domain"],
 757                         "reason" : row["reason"],
 758                     })
 759
 760         logger.debug("Invoking commit() ...")
 761         database.connection.commit()
 762
 763         logger.debug("config.get(bot_enabled)='%s',blockdict()=%d", config.get("bot_enabled"), len(blockdict))
 764         if config.get("bot_enabled") and len(blockdict) > 0:
 765             logger.info("Sending bot POST for blocker='%s',blockdict()=%d ...", blocker, len(blockdict))
 766             network.send_bot_post(blocker, blockdict)
 767
 768     logger.debug("Checking if blocker='%s' has pending updates ...", blocker)
 769     if instances.has_pending(blocker):
 770         logger.debug("Flushing updates for blocker='%s' ...", blocker)
 771         instances.update(blocker)
 772
 773     logger.debug("Success! - EXIT!")
 774     return 0
 775
 776 def fetch_fba_rss(args: argparse.Namespace) -> int:
 777     logger.debug("args[]='%s' - CALLED!", type(args))
 778
 779     domains = list()
 780
 781     logger.debug("Invoking locking.acquire() ...")
 782     locking.acquire()
 783
 784     components = urlparse(args.feed)
 785     domain = components.netloc.lower().split(":")[0]
 786
 787     logger.debug("domain='%s'", domain)
 788     if sources.is_recent(domain):
 789         logger.info("API from domain='%s' has recently being accessed - EXIT!", domain)
 790         return 0
 791     else:
 792         logger.debug("domain='%s' has not been recently used, marking ...", domain)
 793         sources.update(domain)
 794
 795     logger.info("Fetch FBA-specific RSS args.feed='%s' ...", args.feed)
 796     response = utils.fetch_url(args.feed, network.web_headers, (config.get("connection_timeout"), config.get("read_timeout")))
 797
 798     logger.debug("response.ok='%s',response.status_code=%d,response.text()=%d", response.ok, response.status_code, len(response.text))
 799     if response.ok and response.status_code == 200 and len(response.text) > 0:
 800         logger.debug("Parsing RSS feed (%d Bytes) ...", len(response.text))
 801         rss = atoma.parse_rss_bytes(response.content)
 802
 803         logger.debug("rss[]='%s'", type(rss))
 804         for item in rss.items:
 805             logger.debug("item[%s]='%s'", type(item), item)
 806             domain = item.link.split("=")[1]
 807             domain = tidyup.domain(domain) if domain not in[None, ""] else None
 808
 809             logger.debug("domain='%s' - AFTER!", domain)
 810             if domain is None or domain == "":
 811                 logger.debug("domain='%s' is empty after tidyup.domain() - SKIPPED!", domain)
 812                 continue
 813
 814             logger.debug("domain='%s' - BEFORE!", domain)
 815             domain = domain.encode("idna").decode("utf-8")
 816             logger.debug("domain='%s' - AFTER!", domain)
 817
 818             if not domain_helper.is_wanted(domain):
 819                 logger.debug("domain='%s' is not wanted - SKIPPED!", domain)
 820                 continue
 821             elif domain in domains:
 822                 logger.debug("domain='%s' is already added - SKIPPED!", domain)
 823                 continue
 824             elif instances.is_registered(domain):
 825                 logger.debug("domain='%s' is already registered - SKIPPED!", domain)
 826                 continue
 827             elif instances.is_recent(domain):
 828                 logger.debug("domain='%s' has been recently crawled - SKIPPED!", domain)
 829                 continue
 830
 831             logger.debug("Adding domain='%s'", domain)
 832             domains.append(domain)
 833
 834     logger.debug("domains()=%d", len(domains))
 835     if len(domains) > 0:
 836         logger.info("Adding %d new instances ...", len(domains))
 837         for domain in domains:
 838             logger.debug("domain='%s'", domain)
 839             try:
 840                 logger.info("Fetching instances from domain='%s' ...", domain)
 841                 federation.fetch_instances(domain, None, None, inspect.currentframe().f_code.co_name)
 842             except network.exceptions as exception:
 843                 logger.warning("Exception '%s' during fetching instances (fetch_fba_rss) from domain='%s'", type(exception), domain)
 844                 instances.set_last_error(domain, exception)
 845                 return 100
 846
 847     logger.debug("Success! - EXIT!")
 848     return 0
 849
 850 def fetch_fbabot_atom(args: argparse.Namespace) -> int:
 851     logger.debug("args[]='%s' - CALLED!", type(args))
 852
 853     logger.debug("Invoking locking.acquire() ...")
 854     locking.acquire()
 855
 856     source_domain = "ryona.agency"
 857     feed = f"https://{source_domain}/users/fba/feed.atom"
 858
 859     logger.debug("args.feed[%s]='%s'", type(args.feed), args.feed)
 860     if args.feed is not None and validators.url(args.feed):
 861         logger.debug("Setting feed='%s' ...", args.feed)
 862         feed = str(args.feed)
 863         source_domain = urlparse(args.feed).netloc
 864
 865     if sources.is_recent(source_domain):
 866         logger.info("API from source_domain='%s' has recently being accessed - EXIT!", source_domain)
 867         return 1
 868     else:
 869         logger.debug("source_domain='%s' has not been recently used, marking ...", source_domain)
 870         sources.update(source_domain)
 871
 872     domains = list()
 873
 874     logger.info("Fetching ATOM feed='%s' from FBA bot account ...", feed)
 875     response = utils.fetch_url(feed, network.web_headers, (config.get("connection_timeout"), config.get("read_timeout")))
 876
 877     logger.debug("response.ok='%s',response.status_code=%d,response.text()=%d", response.ok, response.status_code, len(response.text))
 878     if response.ok and response.status_code == 200 and len(response.text) > 0:
 879         logger.debug("Parsing ATOM feed (%d Bytes) ...", len(response.text))
 880         atom = atoma.parse_atom_bytes(response.content)
 881
 882         logger.debug("atom[]='%s'", type(atom))
 883         for entry in atom.entries:
 884             logger.debug("entry[]='%s'", type(entry))
 885             doc = bs4.BeautifulSoup(entry.content.value, "html.parser")
 886             logger.debug("doc[]='%s'", type(doc))
 887
 888             for element in doc.findAll("a"):
 889                 logger.debug("element[]='%s'", type(element))
 890                 for href in element["href"].split(","):
 891                     logger.debug("href[%s]='%s' - BEFORE!", type(href), href)
 892                     domain = tidyup.domain(href) if href not in [None, ""] else None
 893
 894                     logger.debug("domain='%s' - AFTER!", domain)
 895                     if domain is None or domain == "":
 896                         logger.debug("domain='%s' is empty after tidyup.domain() - SKIPPED!", domain)
 897                         continue
 898
 899                     logger.debug("domain='%s' - BEFORE!", domain)
 900                     domain = domain.encode("idna").decode("utf-8")
 901                     logger.debug("domain='%s' - AFTER!", domain)
 902
 903                     if not domain_helper.is_wanted(domain):
 904                         logger.debug("domain='%s' is not wanted - SKIPPED!", domain)
 905                         continue
 906                     elif domain in domains:
 907                         logger.debug("domain='%s' is already added - SKIPPED!", domain)
 908                         continue
 909                     elif instances.is_registered(domain):
 910                         logger.debug("domain='%s' is already registered - SKIPPED!", domain)
 911                         continue
 912                     elif instances.is_recent(domain):
 913                         logger.debug("domain='%s' has been recently crawled - SKIPPED!", domain)
 914                         continue
 915
 916                     logger.debug("Adding domain='%s',domains()=%d", domain, len(domains))
 917                     domains.append(domain)
 918
 919     logger.debug("domains()=%d", len(domains))
 920     if len(domains) > 0:
 921         logger.info("Adding %d new instances ...", len(domains))
 922         for domain in domains:
 923             logger.debug("domain='%s'", domain)
 924             try:
 925                 logger.info("Fetching instances from domain='%s' ...", domain)
 926                 federation.fetch_instances(domain, source_domain, None, inspect.currentframe().f_code.co_name)
 927             except network.exceptions as exception:
 928                 logger.warning("Exception '%s' during fetching instances (fetch_fbabot_atom) from domain='%s'", type(exception), domain)
 929                 instances.set_last_error(domain, exception)
 930                 return 100
 931
 932     logger.debug("Success! - EXIT!")
 933     return 0
 934
 935 def fetch_instances(args: argparse.Namespace) -> int:
 936     logger.debug("args[]='%s' - CALLED!", type(args))
 937
 938     logger.debug("args.domain='%s' - checking ...", args.domain)
 939     if not validators.domain(args.domain):
 940         logger.warning("args.domain='%s' is not valid.", args.domain)
 941         return 100
 942     elif blacklist.is_blacklisted(args.domain):
 943         logger.warning("args.domain='%s' is blacklisted, won't check it!", args.domain)
 944         return 101
 945
 946     logger.debug("Invoking locking.acquire() ...")
 947     locking.acquire()
 948
 949     # Initialize values
 950     domain = tidyup.domain(args.domain)
 951     origin = software = None
 952
 953     # Fetch record
 954     database.cursor.execute("SELECT origin, software FROM instances WHERE domain = ? LIMIT 1", [args.domain])
 955     row = database.cursor.fetchone()
 956     if row is not None:
 957         origin = row["origin"]
 958         software = row["software"]
 959
 960     if software_helper.is_relay(software):
 961         logger.warning("args.domain='%s' is of software type '%s' which is not supported by this command. Please invoke fetch_relays instead.", args.domain, software)
 962         return 102
 963
 964     # Initial fetch
 965     try:
 966         logger.info("Fetching instances from args.domain='%s',origin='%s',software='%s' ...", domain, origin, software)
 967         federation.fetch_instances(domain, origin, software, inspect.currentframe().f_code.co_name)
 968     except network.exceptions as exception:
 969         logger.warning("Exception '%s' during fetching instances (fetch_instances) from args.domain='%s'", type(exception), args.domain)
 970         instances.set_last_error(args.domain, exception)
 971         instances.update(args.domain)
 972         return 100
 973
 974     if args.single:
 975         logger.debug("Not fetching more instances - EXIT!")
 976         return 0
 977
 978     # Loop through some instances
 979     database.cursor.execute(
 980         "SELECT domain, origin, software, nodeinfo_url FROM instances WHERE software IN ('pleroma', 'mastodon', 'friendica', 'misskey', 'lemmy', 'peertube', 'takahe', 'gotosocial', 'brighteon', 'wildebeest', 'bookwyrm', 'mitra', 'areionskey', 'mammuthus', 'neodb') AND (last_instance_fetch IS NULL OR last_instance_fetch < ?) ORDER BY total_peers DESC, last_response_time ASC, last_updated ASC", [time.time() - config.get("recheck_instance")]
 981     )
 982
 983     rows = database.cursor.fetchall()
 984     logger.info("Checking %d entries ...", len(rows))
 985     for row in rows:
 986         logger.debug("row[domain]='%s'", row["domain"])
 987         if row["domain"] == "":
 988             logger.debug("row[domain] is empty - SKIPPED!")
 989             continue
 990
 991         logger.debug("row[domain]='%s' - BEFORE!", row["domain"])
 992         domain = row["domain"].encode("idna").decode("utf-8")
 993         logger.debug("domain='%s' - AFTER!", domain)
 994
 995         if not domain_helper.is_wanted(domain):
 996             logger.debug("Domain domain='%s' is not wanted - SKIPPED!", domain)
 997             continue
 998
 999         try:
1000             logger.info("Fetching instances for domain='%s',origin='%s',software='%s',nodeinfo_url='%s'", domain, row["origin"], row["software"], row["nodeinfo_url"])
1001             federation.fetch_instances(domain, row["origin"], row["software"], inspect.currentframe().f_code.co_name, row["nodeinfo_url"])
1002         except network.exceptions as exception:
1003             logger.warning("Exception '%s' during fetching instances (fetch_instances) from domain='%s'", type(exception), domain)
1004             instances.set_last_error(domain, exception)
1005
1006     logger.debug("Success - EXIT!")
1007     return 0
1008
1009 def fetch_csv(args: argparse.Namespace) -> int:
1010     logger.debug("args[]='%s' - CALLED!", type(args))
1011
1012     logger.debug("Invoking locking.acquire() ...")
1013     locking.acquire()
1014
1015     logger.info("Checking %d CSV files ...", len(blocklists.csv_files))
1016     for block in blocklists.csv_files:
1017         logger.debug("block[blocker]='%s',block[csv_url]='%s'", block["blocker"], block["csv_url"])
1018
1019         # Is domain given and not equal blocker?
1020         if isinstance(args.domain, str) and args.domain != block["blocker"]:
1021             logger.debug("Skipping blocker='%s', not matching args.domain='%s'", block["blocker"], args.domain)
1022             continue
1023
1024         logger.debug("Invoking processing.csv_block(%s, %s, fetch_csv) ...", block["blocker"], block["csv_url"])
1025         processing.csv_block(block["blocker"], block["csv_url"], inspect.currentframe().f_code.co_name)
1026
1027     logger.debug("Success - EXIT!")
1028     return 0
1029
1030 def fetch_oliphant(args: argparse.Namespace) -> int:
1031     logger.debug("args[]='%s' - CALLED!", type(args))
1032
1033     logger.debug("Invoking locking.acquire() ...")
1034     locking.acquire()
1035
1036     source_domain = "codeberg.org"
1037     if sources.is_recent(source_domain):
1038         logger.info("API from source_domain='%s' has recently being accessed - EXIT!", source_domain)
1039         return 1
1040     else:
1041         logger.debug("source_domain='%s' has not been recently used, marking ...", source_domain)
1042         sources.update(source_domain)
1043
1044     # Base URL
1045     base_url = f"https://{source_domain}/oliphant/blocklists/raw/branch/main/blocklists"
1046
1047     logger.debug("Downloading %d files ...", len(blocklists.oliphant_blocklists))
1048     for block in blocklists.oliphant_blocklists:
1049         # Is domain given and not equal blocker?
1050         logger.debug("block[blocker]='%s',block[csv_url]='%s'", block["blocker"], block["csv_url"])
1051         if isinstance(args.domain, str) and args.domain != block["blocker"]:
1052             logger.debug("Skipping blocker='%s', not matching args.domain='%s'", block["blocker"], args.domain)
1053             continue
1054
1055         url = f"{base_url}/{block['csv_url']}"
1056
1057         logger.debug("Invoking processing.csv_block(%s, %s, fetch_oliphant) ...", block["blocker"], url)
1058         processing.csv_block(block["blocker"], url, inspect.currentframe().f_code.co_name)
1059
1060     logger.debug("Success! - EXIT!")
1061     return 0
1062
1063 def fetch_txt(args: argparse.Namespace) -> int:
1064     logger.debug("args[]='%s' - CALLED!", type(args))
1065
1066     logger.debug("Invoking locking.acquire() ...")
1067     locking.acquire()
1068
1069     logger.info("Checking %d text file(s) ...", len(blocklists.txt_files))
1070     for row in blocklists.txt_files:
1071         logger.debug("Fetching row[url]='%s' ...", row["url"])
1072         response = utils.fetch_url(row["url"], network.web_headers, (config.get("connection_timeout"), config.get("read_timeout")))
1073
1074         logger.debug("response.ok='%s',response.status_code=%d,response.text()=%d", response.ok, response.status_code, len(response.text))
1075         if response.ok and response.status_code == 200 and response.text != "":
1076             logger.debug("Returned %d Bytes for processing", len(response.text.strip()))
1077             domains = response.text.strip().split("\n")
1078
1079             logger.info("Processing %d domains ...", len(domains))
1080             for domain in domains:
1081                 logger.debug("domain='%s' - BEFORE!", domain)
1082                 domain = tidyup.domain(domain) if domain not in[None, ""] else None
1083
1084                 logger.debug("domain='%s' - AFTER!", domain)
1085                 if domain is None or domain == "":
1086                     logger.debug("domain='%s' is empty after tidyup.domain() - SKIPPED!", domain)
1087                     continue
1088                 elif not domain_helper.is_wanted(domain):
1089                     logger.debug("domain='%s' is not wanted - SKIPPED!", domain)
1090                     continue
1091                 elif instances.is_recent(domain):
1092                     logger.debug("domain='%s' has been recently crawled - SKIPPED!", domain)
1093                     continue
1094
1095                 logger.debug("Processing domain='%s',row[blocker]='%s'", domain, row["blocker"])
1096                 processed = processing.instance(domain, row["blocker"], inspect.currentframe().f_code.co_name)
1097
1098                 logger.debug("processed='%s'", processed)
1099                 if not processed:
1100                     logger.debug("domain='%s' was not generically processed - SKIPPED!", domain)
1101                     continue
1102
1103     logger.debug("Success! - EXIT!")
1104     return 0
1105
1106 def fetch_fedipact(args: argparse.Namespace) -> int:
1107     logger.debug("args[]='%s' - CALLED!", type(args))
1108
1109     logger.debug("Invoking locking.acquire() ...")
1110     locking.acquire()
1111
1112     source_domain = "fedipact.online"
1113     if sources.is_recent(source_domain):
1114         logger.info("API from source_domain='%s' has recently being accessed - EXIT!", source_domain)
1115         return 1
1116     else:
1117         logger.debug("source_domain='%s' has not been recently used, marking ...", source_domain)
1118         sources.update(source_domain)
1119
1120     logger.info("Fetching / from source_domain='%s' ...", source_domain)
1121     response = utils.fetch_url(
1122         f"https://{source_domain}",
1123         network.web_headers,
1124         (config.get("connection_timeout"), config.get("read_timeout"))
1125     )
1126
1127     logger.debug("response.ok='%s',response.status_code=%d,response.text()=%d", response.ok, response.status_code, len(response.text))
1128     if response.ok and response.status_code == 200 and response.text != "":
1129         logger.debug("Parsing %d Bytes ...", len(response.text))
1130
1131         doc = bs4.BeautifulSoup(response.text, "html.parser")
1132         logger.debug("doc[]='%s'", type(doc))
1133
1134         rows = doc.findAll("li")
1135         logger.info("Checking %d row(s) ...", len(rows))
1136         for row in rows:
1137             logger.debug("row[]='%s'", type(row))
1138             domain = tidyup.domain(row.contents[0]) if row.contents[0] not in [None, ""] else None
1139
1140             logger.debug("domain='%s' - AFTER!", domain)
1141             if domain is None or domain == "":
1142                 logger.debug("domain='%s' is empty after tidyup.domain() - SKIPPED!", domain)
1143                 continue
1144
1145             logger.debug("domain='%s' - BEFORE!", domain)
1146             domain = domain.encode("idna").decode("utf-8")
1147             logger.debug("domain='%s' - AFTER!", domain)
1148
1149             if not domain_helper.is_wanted(domain):
1150                 logger.debug("domain='%s' is not wanted - SKIPPED!", domain)
1151                 continue
1152             elif instances.is_registered(domain):
1153                 logger.debug("domain='%s' is already registered - SKIPPED!", domain)
1154                 continue
1155             elif instances.is_recent(domain):
1156                 logger.debug("domain='%s' has been recently crawled - SKIPPED!", domain)
1157                 continue
1158
1159             logger.info("Fetching domain='%s' ...", domain)
1160             federation.fetch_instances(domain, "beach.city", None, inspect.currentframe().f_code.co_name)
1161
1162     logger.debug("Success! - EXIT!")
1163     return 0
1164
1165 def fetch_joinmobilizon(args: argparse.Namespace) -> int:
1166     logger.debug("args[]='%s' - CALLED!", type(args))
1167
1168     logger.debug("Invoking locking.acquire() ...")
1169     locking.acquire()
1170
1171     source_domain = "instances.joinmobilizon.org"
1172     if sources.is_recent(source_domain):
1173         logger.info("API from source_domain='%s' has recently being accessed - EXIT!", source_domain)
1174         return 1
1175     else:
1176         logger.debug("source_domain='%s' has not been recently used, marking ...", source_domain)
1177         sources.update(source_domain)
1178
1179     logger.info("Fetching instances from source_domain='%s' ...", source_domain)
1180     raw = utils.fetch_url(
1181         f"https://{source_domain}/api/v1/instances",
1182         network.web_headers,
1183         (config.get("connection_timeout"), config.get("read_timeout"))
1184     ).text
1185     logger.debug("raw()=%d,raw[]='%s'", len(raw), type(raw))
1186
1187     parsed = json.loads(raw)
1188     logger.debug("parsed[%s]()=%d", type(parsed), len(parsed))
1189
1190     if "data" not in parsed:
1191         logger.warning("parsed()=%d does not contain key 'data'")
1192         return 1
1193
1194     logger.info("Checking %d instances ...", len(parsed["data"]))
1195     for row in parsed["data"]:
1196         logger.debug("row[]='%s'", type(row))
1197         if "host" not in row:
1198             logger.warning("row='%s' does not contain key 'host' - SKIPPED!", row)
1199             continue
1200         elif not domain_helper.is_wanted(row["host"]):
1201             logger.debug("row[host]='%s' is not wanted - SKIPPED!", row["host"])
1202             continue
1203         elif instances.is_registered(row["host"]):
1204             logger.debug("row[host]='%s' is already registered - SKIPPED!", row["host"])
1205             continue
1206
1207         logger.info("Fetching row[host]='%s' ...", row["host"])
1208         federation.fetch_instances(row["host"], "demo.mobilizon.org", None, inspect.currentframe().f_code.co_name)
1209
1210     logger.debug("Success! - EXIT!")
1211     return 0
1212
1213 def fetch_joinmisskey(args: argparse.Namespace) -> int:
1214     logger.debug("args[]='%s' - CALLED!", type(args))
1215
1216     logger.debug("Invoking locking.acquire() ...")
1217     locking.acquire()
1218
1219     source_domain = "instanceapp.misskey.page"
1220     if sources.is_recent(source_domain):
1221         logger.info("API from source_domain='%s' has recently being accessed - EXIT!", source_domain)
1222         return 1
1223     else:
1224         logger.debug("source_domain='%s' has not been recently used, marking ...", source_domain)
1225         sources.update(source_domain)
1226
1227     logger.info("Fetching instances.json from source_domain='%s' ...", source_domain)
1228     raw = utils.fetch_url(
1229         f"https://{source_domain}/instances.json",
1230         network.web_headers,
1231         (config.get("connection_timeout"), config.get("read_timeout"))
1232     ).text
1233     logger.debug("raw()=%d,raw[]='%s'", len(raw), type(raw))
1234
1235     parsed = json.loads(raw)
1236     logger.debug("parsed[%s]()=%d", type(parsed), len(parsed))
1237
1238     if "instancesInfos" not in parsed:
1239         logger.warning("parsed()=%d does not contain element 'instancesInfos'")
1240         return 1
1241
1242     logger.info("Checking %d instane(s) ...", len(parsed["instancesInfos"]))
1243     for row in parsed["instancesInfos"]:
1244         logger.debug("row[%s]='%s'", type(row), row)
1245         if "url" not in row:
1246             logger.warning("row()=%d does not have element 'url' - SKIPPED!", len(row))
1247             continue
1248         elif not domain_helper.is_wanted(row["url"]):
1249             logger.debug("row[url]='%s' is not wanted - SKIPPED!", row["url"])
1250             continue
1251         elif instances.is_registered(row["url"]):
1252             logger.debug("row[url]='%s' is already registered - SKIPPED!", row["url"])
1253             continue
1254
1255         logger.info("Fetching row[url]='%s' ...", row["url"])
1256         federation.fetch_instances(row["url"], "misskey.io", None, inspect.currentframe().f_code.co_name)
1257
1258     logger.debug("Success! - EXIT!")
1259     return 0
1260
1261 def recheck_obfuscation(args: argparse.Namespace) -> int:
1262     logger.debug("args[]='%s' - CALLED!", type(args))
1263
1264     logger.debug("Invoking locking.acquire() ...")
1265     locking.acquire()
1266
1267     if isinstance(args.domain, str) and args.domain != "" and domain_helper.is_wanted(args.domain):
1268         database.cursor.execute("SELECT domain, software, nodeinfo_url FROM instances WHERE (has_obfuscation = 1 OR has_obfuscation IS NULL) AND domain = ?", [args.domain])
1269     elif isinstance(args.software, str) and args.software != "" and validators.domain(args.software) == args.software:
1270         database.cursor.execute("SELECT domain, software, nodeinfo_url FROM instances WHERE (has_obfuscation = 1 OR has_obfuscation IS NULL) AND software = ?", [args.software])
1271     else:
1272         database.cursor.execute("SELECT domain, software, nodeinfo_url FROM instances WHERE has_obfuscation = 1 OR has_obfuscation IS NULL")
1273
1274     rows = database.cursor.fetchall()
1275     logger.info("Checking %d domains ...", len(rows))
1276     for row in rows:
1277         logger.debug("Fetching peers from domain='%s',software='%s',nodeinfo_url='%s' ...", row["domain"], row["software"], row["nodeinfo_url"])
1278         if (args.force is None or not args.force) and args.domain is None and args.software is None and instances.is_recent(row["domain"], "last_blocked"):
1279             logger.debug("row[domain]='%s' has been recently checked, args.force[]='%s' - SKIPPED!", row["domain"], type(args.force))
1280             continue
1281
1282         logger.debug("Invoking federation.fetch_blocks(%s) ...", row["domain"])
1283         blocking = federation.fetch_blocks(row["domain"])
1284
1285         logger.debug("blocking()=%d", len(blocking))
1286         if len(blocking) == 0:
1287             if row["software"] == "pleroma":
1288                 logger.debug("domain='%s',software='%s'", row["domain"], row["software"])
1289                 blocking = pleroma.fetch_blocks(row["domain"])
1290             elif row["software"] == "mastodon":
1291                 logger.debug("domain='%s',software='%s'", row["domain"], row["software"])
1292                 blocking = mastodon.fetch_blocks(row["domain"])
1293             elif row["software"] == "lemmy":
1294                 logger.debug("domain='%s',software='%s'", row["domain"], row["software"])
1295                 blocking = lemmy.fetch_blocks(row["domain"])
1296             elif row["software"] == "friendica":
1297                 logger.debug("domain='%s',software='%s'", row["domain"], row["software"])
1298                 blocking = friendica.fetch_blocks(row["domain"])
1299             elif row["software"] == "misskey":
1300                 logger.debug("domain='%s',software='%s'", row["domain"], row["software"])
1301                 blocking = misskey.fetch_blocks(row["domain"])
1302             else:
1303                 logger.warning("Unknown software: domain='%s',software='%s'", row["domain"], row["software"])
1304
1305         # c.s isn't part of oliphant's "hidden" blocklists
1306         logger.debug("row[domain]='%s'", row["domain"])
1307         if row["domain"] != "chaos.social" and not software_helper.is_relay(row["software"]) and not blocklists.has(row["domain"]):
1308             logger.debug("Invoking instances.set_total_blocks(%s, %d) ...", row["domain"], len(blocking))
1309             instances.set_last_blocked(row["domain"])
1310             instances.set_total_blocks(row["domain"], blocking)
1311
1312         obfuscated = 0
1313         blockdict = list()
1314
1315         logger.info("Checking %d block(s) from domain='%s' ...", len(blocking), row["domain"])
1316         for block in blocking:
1317             logger.debug("block[blocked]='%s'", block["blocked"])
1318             blocked = None
1319
1320             if block["blocked"] == "":
1321                 logger.debug("block[blocked] is empty - SKIPPED!")
1322                 continue
1323             elif block["blocked"].endswith(".onion"):
1324                 logger.debug("blocked='%s' is a TOR onion domain name - SKIPPED!", block["blocked"])
1325                 continue
1326             elif block["blocked"].endswith(".i2p") and config.get("allow_i2p_domain") == "true":
1327                 logger.debug("blocked='%s' is an I2P onion domain name - SKIPPED!", block["blocked"])
1328                 continue
1329             elif block["blocked"].endswith(".arpa"):
1330                 logger.debug("blocked='%s' is a reversed IP address - SKIPPED!", block["blocked"])
1331                 continue
1332             elif block["blocked"].endswith(".tld"):
1333                 logger.debug("blocked='%s' is a fake domain name - SKIPPED!", block["blocked"])
1334                 continue
1335             elif block["blocked"].find("*") >= 0 or block["blocked"].find("?") >= 0:
1336                 logger.debug("block='%s' is obfuscated.", block["blocked"])
1337                 obfuscated = obfuscated + 1
1338                 blocked = utils.deobfuscate(block["blocked"], row["domain"], block["digest"] if "digest" in block else None)
1339             elif not domain_helper.is_wanted(block["blocked"]):
1340                 logger.debug("blocked='%s' is not wanted - SKIPPED!", block["blocked"])
1341                 continue
1342             elif blocks.is_instance_blocked(row["domain"], block["blocked"]):
1343                 logger.debug("blocked='%s' is already blocked - SKIPPED!", block["blocked"])
1344                 continue
1345
1346             logger.debug("blocked[%s]='%s',block[blocked]='%s'", type(blocked), blocked, block["blocked"])
1347             if blocked is not None and blocked != block["blocked"]:
1348                 logger.debug("blocked='%s' was deobfuscated to blocked='%s'", block["blocked"], blocked)
1349                 obfuscated = obfuscated - 1
1350
1351                 if blacklist.is_blacklisted(blocked):
1352                     logger.debug("blocked='%s' is blacklisted - SKIPPED!", blocked)
1353                     continue
1354                 elif blacklist.is_blacklisted(row["domain"]):
1355                     logger.debug("row[domain]='%s' is blacklisted - SKIPPED!", row["domain"])
1356                     continue
1357                 elif blocks.is_instance_blocked(row["domain"], blocked):
1358                     logger.debug("blocked='%s' is already blocked by domain='%s' - SKIPPED!", blocked, row["domain"])
1359                     continue
1360
1361                 block["block_level"] = blocks.alias_block_level(block["block_level"])
1362
1363                 logger.info("blocked='%s' has been deobfuscated to blocked='%s', adding ...", block["blocked"], blocked)
1364                 if processing.block(row["domain"], blocked, block["reason"], block["block_level"]) and block["block_level"] == "reject" and config.get("bot_enabled"):
1365                     logger.debug("Appending blocked='%s',reason='%s' for blocker='%s' ...", block["blocked"], block["block_level"], row["domain"])
1366                     blockdict.append({
1367                         "blocked": blocked,
1368                         "reason" : block["reason"],
1369                     })
1370
1371         logger.debug("Setting obfuscated=%d for row[domain]='%s' ...", obfuscated, row["domain"])
1372         instances.set_has_obfuscation(row["domain"], (obfuscated > 0))
1373         instances.set_obfuscated_blocks(row["domain"], obfuscated)
1374
1375         logger.info("domain='%s' has %d obfuscated domain(s)", row["domain"], obfuscated)
1376         if instances.has_pending(row["domain"]):
1377             logger.debug("Flushing updates for blocker='%s' ...", row["domain"])
1378             instances.update(row["domain"])
1379
1380         logger.debug("Invoking commit() ...")
1381         database.connection.commit()
1382
1383         logger.debug("config.get(bot_enabled)='%s',blockdict()=%d", config.get("bot_enabled"), len(blockdict))
1384         if config.get("bot_enabled") and len(blockdict) > 0:
1385             logger.info("Sending bot POST for blocker='%s,blockdict()=%d ...", row["domain"], len(blockdict))
1386             network.send_bot_post(row["domain"], blockdict)
1387
1388     logger.debug("Success! - EXIT!")
1389     return 0
1390
1391 def fetch_fedilist(args: argparse.Namespace) -> int:
1392     logger.debug("args[]='%s' - CALLED!", type(args))
1393
1394     logger.debug("Invoking locking.acquire() ...")
1395     locking.acquire()
1396
1397     source_domain = "demo.fedilist.com"
1398     if sources.is_recent(source_domain):
1399         logger.info("API from source_domain='%s' has recently being accessed - EXIT!", source_domain)
1400         return 1
1401     else:
1402         logger.debug("source_domain='%s' has not been recently used, marking ...", source_domain)
1403         sources.update(source_domain)
1404
1405     url = f"http://{source_domain}/instance/csv?onion=not"
1406     if args.software is not None and args.software != "":
1407         logger.debug("args.software='%s'", args.software)
1408         url = f"http://{source_domain}/instance/csv?software={args.software}&onion=not"
1409
1410     logger.info("Fetching url='%s' ...", url)
1411     response = reqto.get(
1412         url,
1413         headers=network.web_headers,
1414         timeout=(config.get("connection_timeout"), config.get("read_timeout")),
1415         allow_redirects=False
1416     )
1417
1418     logger.debug("response.ok='%s',response.status_code=%d,response.text()=%d", response.ok, response.status_code, len(response.text))
1419     if not response.ok or response.status_code > 200 or len(response.content) == 0:
1420         logger.warning("Failed fetching url='%s': response.ok='%s',response.status_code=%d,response.content()=%d - EXIT!", url, response.ok, response.status_code, len(response.text))
1421         return 1
1422
1423     reader = csv.DictReader(response.content.decode("utf-8").splitlines(), dialect="unix")
1424
1425     logger.debug("reader[]='%s'", type(reader))
1426     if reader is None:
1427         logger.warning("Failed parsing response.content()=%d as CSV content", len(response.content))
1428         return 2
1429
1430     rows = list(reader)
1431
1432     logger.info("Checking %d rows ...", len(rows))
1433     for row in rows:
1434         logger.debug("row[]='%s'", type(row))
1435         if "hostname" not in row:
1436             logger.warning("row()=%d has no element 'hostname' - SKIPPED!", len(row))
1437             continue
1438
1439         logger.debug("row[hostname]='%s' - BEFORE!", row["hostname"])
1440         domain = tidyup.domain(row["hostname"]) if row["hostname"] not in [None, ""] else None
1441         logger.debug("domain='%s' - AFTER!", domain)
1442
1443         if domain is None or domain == "":
1444             logger.debug("domain='%s' is empty after tidyup.domain(): row[hostname]='%s' - SKIPPED!", domain, row["hostname"])
1445             continue
1446
1447         logger.debug("domain='%s' - BEFORE!", domain)
1448         domain = domain.encode("idna").decode("utf-8")
1449         logger.debug("domain='%s' - AFTER!", domain)
1450
1451         if not domain_helper.is_wanted(domain):
1452             logger.debug("domain='%s' is not wanted - SKIPPED!", domain)
1453             continue
1454         elif (args.force is None or not args.force) and instances.is_registered(domain):
1455             logger.debug("domain='%s' is already registered, --force not specified: args.force[]='%s'", domain, type(args.force))
1456             continue
1457         elif instances.is_recent(domain):
1458             logger.debug("domain='%s' has been recently crawled - SKIPPED!", domain)
1459             continue
1460
1461         logger.info("Fetching instances from domain='%s' ...", domain)
1462         federation.fetch_instances(domain, None, None, inspect.currentframe().f_code.co_name)
1463
1464     logger.debug("Success! - EXIT!")
1465     return 0
1466
1467 def update_nodeinfo(args: argparse.Namespace) -> int:
1468     logger.debug("args[]='%s' - CALLED!", type(args))
1469
1470     logger.debug("Invoking locking.acquire() ...")
1471     locking.acquire()
1472
1473     if args.domain is not None and args.domain != "":
1474         logger.debug("Fetching args.domain='%s'", args.domain)
1475         database.cursor.execute("SELECT domain, software FROM instances WHERE domain = ? LIMIT 1", [args.domain])
1476     elif args.software is not None and args.software != "":
1477         logger.info("Fetching domains for args.software='%s'", args.software)
1478         database.cursor.execute("SELECT domain, software FROM instances WHERE software = ? ORDER BY last_updated ASC", [args.software])
1479     elif args.mode is not None and args.mode != "":
1480         logger.info("Fetching domains for args.mode='%s'", args.mode.upper())
1481         database.cursor.execute("SELECT domain, software FROM instances WHERE detection_mode = ? ORDER BY last_updated ASC", [args.mode])
1482     elif args.no_software:
1483         logger.info("Fetching domains with no software type detected ...")
1484         database.cursor.execute("SELECT domain, software FROM instances WHERE software IS NULL ORDER BY last_updated ASC")
1485     elif args.no_auto:
1486         logger.info("Fetching domains with other detection mode than AUTO_DISOVERY being set ...")
1487         database.cursor.execute("SELECT domain, software FROM instances WHERE detection_mode IS NOT NULL AND detection_mode != 'AUTO_DISCOVERY' ORDER BY last_updated ASC")
1488     elif args.no_detection:
1489         logger.info("Fetching domains with no detection mode being set ...")
1490         database.cursor.execute("SELECT domain, software FROM instances WHERE detection_mode IS NULL ORDER BY last_updated ASC")
1491     else:
1492         logger.info("Fetching domains for recently updated ...")
1493         database.cursor.execute("SELECT domain, software FROM instances ORDER BY last_updated ASC")
1494
1495     domains = database.cursor.fetchall()
1496
1497     logger.info("Checking %d domain(s) ...", len(domains))
1498     cnt = 0
1499     for row in domains:
1500         logger.debug("row[]='%s'", type(row))
1501         if not args.force and instances.is_recent(row["domain"], "last_nodeinfo"):
1502             logger.debug("row[domain]='%s' has been recently checked - SKIPPED!", row["domain"])
1503             continue
1504
1505         try:
1506             logger.info("Checking nodeinfo for row[domain]='%s',row[software]='%s' (%s%%) ...", row["domain"], row["software"], "{:5.1f}".format(cnt / len(domains) * 100))
1507             software = federation.determine_software(row["domain"])
1508
1509             logger.debug("Determined software='%s'", software)
1510             if (software != row["software"] and software is not None) or args.force is True:
1511                 logger.debug("software='%s'", software)
1512                 if software is None:
1513                     logger.debug("Setting nodeinfo_url to 'None' for row[domain]='%s' ...", row["domain"])
1514                     instances.set_nodeinfo_url(row["domain"], None)
1515
1516                 logger.warning("Software type for row[domain]='%s' has changed from '%s' to '%s'!", row["domain"], row["software"], software)
1517                 instances.set_software(row["domain"], software)
1518
1519             if software is not None:
1520                 logger.debug("Setting row[domain]='%s' as successfully determined ...", row["domain"])
1521                 instances.set_success(row["domain"])
1522         except network.exceptions as exception:
1523             logger.warning("Exception '%s' during updating nodeinfo for row[domain]='%s'", type(exception), row["domain"])
1524             instances.set_last_error(row["domain"], exception)
1525
1526         instances.set_last_nodeinfo(row["domain"])
1527         instances.update(row["domain"])
1528         cnt = cnt + 1
1529
1530     logger.debug("Success! - EXIT!")
1531     return 0
1532
1533 def fetch_instances_social(args: argparse.Namespace) -> int:
1534     logger.debug("args[]='%s' - CALLED!", type(args))
1535
1536     logger.debug("Invoking locking.acquire() ...")
1537     locking.acquire()
1538
1539     source_domain = "instances.social"
1540
1541     if config.get("instances_social_api_key") == "":
1542         logger.error("API key not set. Please set in your config.json file.")
1543         return 1
1544     elif sources.is_recent(source_domain):
1545         logger.info("API from source_domain='%s' has recently being accessed - EXIT!", source_domain)
1546         return 2
1547     else:
1548         logger.debug("source_domain='%s' has not been recently used, marking ...", source_domain)
1549         sources.update(source_domain)
1550
1551     headers = {
1552         "Authorization": f"Bearer {config.get('instances_social_api_key')}",
1553     }
1554
1555     logger.info("Fetching list from source_domain='%s' ...", source_domain)
1556     fetched = network.get_json_api(
1557         source_domain,
1558         "/api/1.0/instances/list?count=0&sort_by=name",
1559         headers=headers,
1560         timeout=(config.get("connection_timeout"), config.get("read_timeout"))
1561     )
1562     logger.debug("fetched[]='%s'", type(fetched))
1563
1564     if "error_message" in fetched:
1565         logger.warning("Error during fetching API result: '%s' - EXIT!", fetched["error_message"])
1566         return 2
1567     elif "exception" in fetched:
1568         logger.warning("Exception '%s' during fetching API result - EXIT!", type(fetched["exception"]))
1569         return 3
1570     elif "json" not in fetched:
1571         logger.warning("fetched has no element 'json' - EXIT!")
1572         return 4
1573     elif "instances" not in fetched["json"]:
1574         logger.warning("fetched[row] has no element 'instances' - EXIT!")
1575         return 5
1576
1577     domains = list()
1578     rows = fetched["json"]["instances"]
1579
1580     logger.info("Checking %d row(s) ...", len(rows))
1581     for row in rows:
1582         logger.debug("row[]='%s'", type(row))
1583         domain = tidyup.domain(row["name"]) if row["name"] not in [None, ""] else None
1584         logger.debug("domain='%s' - AFTER!", domain)
1585
1586         if domain is None and domain == "":
1587             logger.debug("domain='%s' is empty after tidyup.domain() - SKIPPED!", domain)
1588             continue
1589
1590         logger.debug("domain='%s' - BEFORE!", domain)
1591         domain = domain.encode("idna").decode("utf-8")
1592         logger.debug("domain='%s' - AFTER!", domain)
1593
1594         if not domain_helper.is_wanted(domain):
1595             logger.debug("domain='%s' is not wanted - SKIPPED!", domain)
1596             continue
1597         elif domain in domains:
1598             logger.debug("domain='%s' is already added - SKIPPED!", domain)
1599             continue
1600         elif instances.is_registered(domain):
1601             logger.debug("domain='%s' is already registered - SKIPPED!", domain)
1602             continue
1603         elif instances.is_recent(domain):
1604             logger.debug("domain='%s' has been recently crawled - SKIPPED!", domain)
1605             continue
1606
1607         logger.info("Fetching instances from domain='%s'", domain)
1608         federation.fetch_instances(domain, None, None, inspect.currentframe().f_code.co_name)
1609
1610     logger.debug("Success! - EXIT!")
1611     return 0
1612
1613 def fetch_relaylist(args: argparse.Namespace) -> int:
1614     logger.debug("args[]='%s' - CALLED!", type(args))
1615
1616     logger.debug("Invoking locking.acquire() ...")
1617     locking.acquire()
1618
1619     source_domain = "api.relaylist.com"
1620
1621     if sources.is_recent(source_domain):
1622         logger.info("API from source_domain='%s' has recently being accessed - EXIT!", source_domain)
1623         return 1
1624     else:
1625         logger.debug("source_domain='%s' has not been recently used, marking ...", source_domain)
1626         sources.update(source_domain)
1627
1628     logger.info("Fetching list from source_domain='%s' ...", source_domain)
1629     fetched = network.get_json_api(
1630         source_domain,
1631         "/relays",
1632         {},
1633         (config.get("connection_timeout"), config.get("read_timeout"))
1634     )
1635     logger.debug("fetched[]='%s'", type(fetched))
1636
1637     if "error_message" in fetched:
1638         logger.warning("Error during fetching API result: '%s' - EXIT!", fetched["error_message"])
1639         return 2
1640     elif "exception" in fetched:
1641         logger.warning("Exception '%s' during fetching API result - EXIT!", type(fetched["exception"]))
1642         return 3
1643     elif "json" not in fetched:
1644         logger.warning("fetched has no element 'json' - EXIT!")
1645         return 4
1646
1647     domains = list()
1648
1649     logger.info("Checking %d row(s) ...", len(fetched["json"]))
1650     for row in fetched["json"]:
1651         logger.debug("row[]='%s'", type(row))
1652         domain = urlparse(row["url"]).netloc.lower().split(":")[0]
1653         logger.debug("domain='%s' - AFTER!", domain)
1654
1655         if domain is None and domain == "":
1656             logger.debug("domain='%s' is empty after tidyup.domain() - SKIPPED!", domain)
1657             continue
1658
1659         logger.debug("domain='%s' - BEFORE!", domain)
1660         domain = domain.encode("idna").decode("utf-8")
1661         logger.debug("domain='%s' - AFTER!", domain)
1662
1663         if not domain_helper.is_wanted(domain):
1664             logger.debug("domain='%s' is not wanted - SKIPPED!", domain)
1665             continue
1666         elif domain in domains:
1667             logger.debug("domain='%s' is already added - SKIPPED!", domain)
1668             continue
1669         elif instances.is_registered(domain):
1670             logger.debug("domain='%s' is already registered - SKIPPED!", domain)
1671             continue
1672         elif instances.is_recent(domain):
1673             logger.debug("domain='%s' has been recently crawled - SKIPPED!", domain)
1674             continue
1675
1676         logger.info("Fetching instances from domain='%s'", domain)
1677         federation.fetch_instances(domain, None, None, inspect.currentframe().f_code.co_name)
1678
1679     logger.debug("Success! - EXIT!")
1680     return 0
1681
1682 def fetch_relays(args: argparse.Namespace) -> int:
1683     logger.debug("args[]='%s' - CALLED!", type(args))
1684
1685     logger.debug("Invoking locking.acquire() ...")
1686     locking.acquire()
1687
1688     if args.domain is not None and args.domain != "":
1689         database.cursor.execute("SELECT domain, software, nodeinfo_url FROM instances WHERE software IN ('activityrelay', 'aoderelay', 'selective-relay', 'pub-relay') AND domain = ? LIMIT 1", [args.domain])
1690     elif args.software is not None and args.software != "":
1691         database.cursor.execute("SELECT domain, software, nodeinfo_url FROM instances WHERE software IN ('activityrelay', 'aoderelay', 'selective-relay', 'pub-relay') AND software = ?", [args.software])
1692     else:
1693         database.cursor.execute("SELECT domain, software, nodeinfo_url FROM instances WHERE software IN ('activityrelay', 'aoderelay', 'selective-relay', 'pub-relay')")
1694
1695     domains = list()
1696     rows = database.cursor.fetchall()
1697
1698     logger.info("Checking %d relays ...", len(rows))
1699     for row in rows:
1700         logger.debug("row[domain]='%s',row[software]='%s' ...", row["domain"], row["software"])
1701         peers = list()
1702         if not args.force and instances.is_recent(row["domain"]):
1703             logger.debug("row[domain]='%s' has been recently fetched - SKIPPED!", row["domain"])
1704             continue
1705
1706         try:
1707             if row["software"] == "pub-relay":
1708                 logger.info("Fetching row[nodeinfo_url]='%s' from relay row[domain]='%s',row[software]='%s' ...", row["nodeinfo_url"], row["domain"], row["software"])
1709                 raw = network.fetch_api_url(
1710                     row["nodeinfo_url"],
1711                     (config.get("connection_timeout"), config.get("read_timeout"))
1712                 )
1713
1714                 logger.debug("raw[%s]()=%d", type(raw), len(raw))
1715                 if "exception" in raw:
1716                     logger.warning("row[domain]='%s' has caused an exception: '%s' - raising again ...", row["domain"], type(raw["exception"]))
1717                     raise raw["exception"]
1718                 elif "error_message" in raw:
1719                     logger.warning("row[domain]='%s' has caused error message: '%s' - SKIPPED!", row["domain"], raw["error_message"])
1720                     instances.set_last_error(row["domain"], raw)
1721                     instances.set_last_instance_fetch(row["domain"])
1722                     instances.update(row["domain"])
1723                     continue
1724                 elif "json" not in raw:
1725                     logger.warning("raw()=%d does not contain key 'json' in response - SKIPPED!", len(raw))
1726                     continue
1727                 elif not "metadata" in raw["json"]:
1728                     logger.warning("raw[json]()=%d does not contain key 'json' in response - SKIPPED!", len(raw["json"]))
1729                     continue
1730                 elif not "peers" in raw["json"]["metadata"]:
1731                     logger.warning("raw[json][metadata()=%d does not contain key 'json' in response - SKIPPED!", len(raw["json"]["metadata"]))
1732                     continue
1733             else:
1734                 logger.info("Fetching / from relay row[domain]='%s',row[software]='%s' ...", row["domain"], row["software"])
1735                 raw = utils.fetch_url(
1736                     f"https://{row['domain']}",
1737                     network.web_headers,
1738                     (config.get("connection_timeout"), config.get("read_timeout"))
1739                 ).text
1740                 logger.debug("raw[%s]()=%d", type(raw), len(raw))
1741
1742                 doc = bs4.BeautifulSoup(raw, features="html.parser")
1743                 logger.debug("doc[]='%s'", type(doc))
1744
1745         except network.exceptions as exception:
1746             logger.warning("Exception '%s' during fetching from relay '%s': '%s'", type(exception), row["domain"], str(exception))
1747             instances.set_last_error(row["domain"], exception)
1748             instances.set_last_instance_fetch(row["domain"])
1749             instances.update(row["domain"])
1750             continue
1751
1752         logger.debug("row[software]='%s'", row["software"])
1753         if row["software"] == "activityrelay":
1754             logger.debug("Checking row[domain]='%s' ...", row["domain"])
1755             tags = doc.findAll("p")
1756
1757             logger.debug("Checking %d paragraphs ...", len(tags))
1758             for tag in tags:
1759                 logger.debug("tag[]='%s'", type(tag))
1760                 if len(tag.contents) == 0:
1761                     logger.debug("tag='%s' is an empty tag - SKIPPED!", tag)
1762                     continue
1763                 elif "registered instances" not in tag.contents[0]:
1764                     logger.debug("Skipping paragraph, text not found.")
1765                     continue
1766
1767                 logger.debug("Found tag.contents[0][]='%s'", tag.contents[0])
1768                 for domain in tag.contents:
1769                     logger.debug("domain[%s]='%s'", type(domain), domain)
1770                     if not isinstance(domain, bs4.element.NavigableString) or "registered instances" in domain:
1771                         continue
1772
1773                     domain = str(domain)
1774                     logger.debug("domain='%s'", domain)
1775                     if not domain_helper.is_wanted(domain):
1776                         logger.debug("domain='%s' is not wanted - SKIPPED!", domain)
1777                         continue
1778
1779                     logger.debug("domain='%s' - BEFORE!", domain)
1780                     domain = tidyup.domain(domain) if domain not in[None, ""] else None
1781                     logger.debug("domain='%s' - AFTER!", domain)
1782
1783                     if domain is None or domain == "":
1784                         logger.debug("domain='%s' is empty after tidyup.domain() from origin='%s' - SKIPPED!", domain, row["domain"])
1785                         continue
1786                     elif domain not in peers:
1787                         logger.debug("Appending domain='%s' to peers list for relay='%s' ...", domain, row["domain"])
1788                         peers.append(domain)
1789
1790                     logger.debug("domains()=%d,domain='%s'", len(domains), domain)
1791                     if dict_helper.has_key(domains, "domain", domain):
1792                         logger.debug("domain='%s' already added", domain)
1793                         continue
1794
1795                     logger.debug("Appending domain='%s',origin='%s',software='%s' ...", domain, row["domain"], row["software"])
1796                     domains.append({
1797                         "domain": domain,
1798                         "origin": row["domain"],
1799                     })
1800         elif row["software"] in ["aoderelay", "selective-relay"]:
1801             logger.debug("Checking row[domain]='%s' ...", row["domain"])
1802             if row["software"] == "aoderelay":
1803                 tags = doc.findAll("section", {"class": "instance"})
1804             else:
1805                 tags = doc.find("div", {"id": "instances"}).findAll("li")
1806
1807             logger.debug("Checking %d tags ...", len(tags))
1808             for tag in tags:
1809                 logger.debug("tag[]='%s'", type(tag))
1810
1811                 link = tag.find("a")
1812                 logger.debug("link[%s]='%s'", type(link), link)
1813                 if not isinstance(link, bs4.element.Tag):
1814                     logger.warning("tag[%s]='%s' is not type of 'bs4.element.Tag' - SKIPPED!", type(tag), tag)
1815                     continue
1816
1817                 components = urlparse(link.get("href"))
1818                 logger.debug("components(%d)='%s'", len(components), components)
1819                 domain = components.netloc.lower().split(":")[0]
1820
1821                 logger.debug("domain='%s' - BEFORE!", domain)
1822                 domain = tidyup.domain(domain) if domain not in[None, ""] else None
1823                 logger.debug("domain='%s' - AFTER!", domain)
1824
1825                 if domain is None or domain == "":
1826                     logger.debug("domain='%s' is empty after tidyup.domain() from origin='%s' - SKIPPED!", domain, row["domain"])
1827                     continue
1828                 elif domain not in peers:
1829                     logger.debug("Appending domain='%s' to peers list for relay='%s' ...", domain, row["domain"])
1830                     peers.append(domain)
1831
1832                 logger.debug("domains()=%d,domain='%s'", len(domains), domain)
1833                 if dict_helper.has_key(domains, "domain", domain):
1834                     logger.debug("domain='%s' already added", domain)
1835                     continue
1836
1837                 logger.debug("Appending domain='%s',origin='%s',software='%s'", domain, row["domain"], row["software"])
1838                 domains.append({
1839                     "domain": domain,
1840                     "origin": row["domain"],
1841                 })
1842         elif row["software"] == "pub-relay":
1843             logger.debug("Checking %d peer(s) row[domain]='%s' ...", len(raw["json"]["metadata"]["peers"]), row["domain"])
1844             for domain in raw["json"]["metadata"]["peers"]:
1845                 logger.debug("domain='%s' - BEFORE!", domain)
1846                 domain = tidyup.domain(domain) if domain not in[None, ""] else None
1847                 logger.debug("domain='%s' - AFTER!", domain)
1848
1849                 if domain is None or domain == "":
1850                     logger.debug("domain='%s' is empty after tidyup.domain() from origin='%s' - SKIPPED!", domain, row["domain"])
1851                     continue
1852                 elif domain not in peers:
1853                     logger.debug("Appending domain='%s' to peers list for relay='%s' ...", domain, row["domain"])
1854                     peers.append(domain)
1855
1856                 logger.debug("domains()=%d,domain='%s'", len(domains), domain)
1857                 if dict_helper.has_key(domains, "domain", domain):
1858                     logger.debug("domain='%s' already added", domain)
1859                     continue
1860
1861                 logger.debug("Appending domain='%s',origin='%s',software='%s' ...", domain, row["domain"], row["software"])
1862                 domains.append({
1863                     "domain": domain,
1864                     "origin": row["domain"],
1865                 })
1866         else:
1867             logger.warning("row[domain]='%s',row[software]='%s' is not supported", row["domain"], row["software"])
1868             continue
1869
1870         logger.debug("Updating last_instance_fetch for row[domain]='%s' ...", row["domain"])
1871         instances.set_last_instance_fetch(row["domain"])
1872
1873         logger.info("Relay '%s' has %d peer(s) registered.", row["domain"], len(peers))
1874         instances.set_total_peers(row["domain"], peers)
1875
1876         logger.debug("Flushing data for row[domain]='%s'", row["domain"])
1877         instances.update(row["domain"])
1878
1879     logger.info("Checking %d domains ...", len(domains))
1880     for row in domains:
1881         logger.debug("row[domain]='%s',row[origin]='%s'", row["domain"], row["origin"])
1882         if not domain_helper.is_wanted(row["domain"]):
1883             logger.debug("row[domain]='%s' is not wanted - SKIPPED!", row["domain"])
1884             continue
1885         elif instances.is_registered(row["domain"]):
1886             logger.debug("row[domain]='%s' is already registered - SKIPPED!", row["domain"])
1887             continue
1888
1889         logger.info("Fetching row[domain]='%s',row[origin]='%s' ...", row["domain"], row["origin"])
1890         federation.fetch_instances(row["domain"], row["origin"], None, inspect.currentframe().f_code.co_name)
1891
1892     logger.debug("Success! - EXIT!")
1893     return 0
1894
1895 def convert_idna(args: argparse.Namespace) -> int:
1896     logger.debug("args[]='%s' - CALLED!", type(args))
1897
1898     database.cursor.execute("SELECT domain FROM instances WHERE domain NOT LIKE '%xn--%' ORDER BY domain ASC")
1899     rows = database.cursor.fetchall()
1900
1901     logger.debug("rows[]='%s'", type(rows))
1902     instances.translate_idnas(rows, "domain")
1903
1904     database.cursor.execute("SELECT origin FROM instances WHERE origin NOT LIKE '%xn--%' ORDER BY origin ASC")
1905     rows = database.cursor.fetchall()
1906
1907     logger.debug("rows[]='%s'", type(rows))
1908     instances.translate_idnas(rows, "origin")
1909
1910     database.cursor.execute("SELECT blocker FROM blocks WHERE blocker NOT LIKE '%xn--%' ORDER BY blocker ASC")
1911     rows = database.cursor.fetchall()
1912
1913     logger.debug("rows[]='%s'", type(rows))
1914     blocks.translate_idnas(rows, "blocker")
1915
1916     database.cursor.execute("SELECT blocked FROM blocks WHERE blocked NOT LIKE '%xn--%' ORDER BY blocked ASC")
1917     rows = database.cursor.fetchall()
1918
1919     logger.debug("rows[]='%s'", type(rows))
1920     blocks.translate_idnas(rows, "blocked")
1921
1922     logger.debug("Success! - EXIT!")
1923     return 0
1924
1925 def remove_invalid(args: argparse.Namespace) -> int:
1926     logger.debug("args[]='%s' - CALLED!", type(args))
1927
1928     logger.debug("Invoking locking.acquire() ...")
1929     locking.acquire()
1930
1931     database.cursor.execute("SELECT domain FROM instances ORDER BY domain ASC")
1932     rows = database.cursor.fetchall()
1933
1934     logger.info("Checking %d domains ...", len(rows))
1935     for row in rows:
1936         logger.debug("row[domain]='%s'", row["domain"])
1937         if not validators.domain(row["domain"].split("/")[0]):
1938             logger.info("Invalid row[domain]='%s' found, removing ...", row["domain"])
1939             database.cursor.execute("DELETE FROM blocks WHERE blocker = ? OR blocked = ?", [row["domain"], row["domain"]])
1940             database.cursor.execute("DELETE FROM instances WHERE domain = ? LIMIT 1", [row["domain"]])
1941
1942     logger.debug("Invoking commit() ...")
1943     database.connection.commit()
1944
1945     logger.info("Vaccum cleaning database ...")
1946     database.cursor.execute("VACUUM")
1947
1948     logger.debug("Success! - EXIT!")
1949     return 0