fba/networks/pleroma.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 logging
  18
  19 import bs4
  20
  21 from fba import database
  22 from fba import utils
  23
  24 from fba.helpers import config
  25 from fba.helpers import domain as domain_helper
  26 from fba.helpers import tidyup
  27
  28 from fba.http import federation
  29 from fba.http import network
  30
  31 from fba.models import instances
  32
  33 logging.basicConfig(level=logging.INFO)
  34 logger = logging.getLogger(__name__)
  35
  36 # Language mapping X -> English
  37 language_mapping = {
  38     # English -> English
  39     "Reject": "Suspended servers",
  40 }
  41
  42 def fetch_blocks(domain: str, nodeinfo_url: str) -> list:
  43     logger.debug("domain='%s',nodeinfo_url='%s' - CALLED!", domain, nodeinfo_url)
  44     domain_helper.raise_on(domain)
  45
  46     if not isinstance(nodeinfo_url, str):
  47         raise ValueError(f"Parameter nodeinfo_url[]='{type(nodeinfo_url)}' is not 'str'")
  48     elif nodeinfo_url == "":
  49         raise ValueError("Parameter 'nodeinfo_url' is empty")
  50
  51     blockdict = list()
  52     rows = None
  53     try:
  54         logger.debug("Fetching nodeinfo: domain='%s',nodeinfo_url='%s'", domain, nodeinfo_url)
  55         rows = federation.fetch_nodeinfo(domain, nodeinfo_url)
  56     except network.exceptions as exception:
  57         logger.warning("Exception '%s' during fetching nodeinfo from domain='%s'", type(exception), domain)
  58         instances.set_last_error(domain, exception)
  59
  60     if rows is None:
  61         logger.warning("Could not fetch nodeinfo from domain='%s'", domain)
  62         return list()
  63     elif "metadata" not in rows:
  64         logger.warning("rows()=%d does not have key 'metadata', domain='%s'", len(rows), domain)
  65         return list()
  66     elif "federation" not in rows["metadata"]:
  67         logger.warning("rows()=%d does not have key 'federation', domain='%s'", len(rows["metadata"]), domain)
  68         return list()
  69
  70     data = rows["metadata"]["federation"]
  71     found = False
  72
  73     logger.debug("data[]='%s'", type(data))
  74     if "mrf_simple" in data:
  75         logger.debug("Found mrf_simple in API response from domain='%s'", domain)
  76         found = True
  77         for block_level, blocklist in (
  78             {
  79                 **data["mrf_simple"],
  80                 **{
  81                     "quarantined_instances": data["quarantined_instances"]
  82                 }
  83             }
  84         ).items():
  85             logger.debug("block_level='%s', blocklist()=%d", block_level, len(blocklist))
  86             block_level = tidyup.domain(block_level)
  87             logger.debug("block_level='%s' - AFTER!", block_level)
  88
  89             if block_level == "":
  90                 logger.warning("block_level is now empty!")
  91                 continue
  92             elif block_level == "accept":
  93                 logger.debug("domain='%s' skipping block_level='accept'", domain)
  94                 continue
  95
  96             block_level = utils.alias_block_level(block_level)
  97
  98             logger.debug("Checking %d entries from domain='%s',block_level='%s' ...", len(blocklist), domain, block_level)
  99             if len(blocklist) > 0:
 100                 for blocked in blocklist:
 101                     logger.debug("blocked='%s' - BEFORE!", blocked)
 102                     blocked = tidyup.domain(blocked)
 103                     logger.debug("blocked='%s' - AFTER!", blocked)
 104
 105                     if blocked == "":
 106                         logger.warning("blocked is empty after tidyup.domain(): domain='%s',block_level='%s'", domain, block_level)
 107                         continue
 108                     elif not utils.is_domain_wanted(blocked):
 109                         logger.debug("blocked='%s' is not wanted - SKIPPED!", blocked)
 110                         continue
 111
 112                     logger.debug("Invoking utils.deobfuscate_domain(%s, %s) ...", blocked, domain)
 113                     blocked = utils.deobfuscate_domain(blocked, domain)
 114
 115                     logger.debug("blocked='%s' - DEOBFUSCATED!", blocked)
 116                     if not utils.is_domain_wanted(blocked):
 117                         logger.debug("blocked='%s' is not wanted - SKIPPED!", blocked)
 118                         continue
 119
 120                     logger.debug("Appending blocker='%s',blocked='%s',block_level='%s' ...", domain, blocked, block_level)
 121                     blockdict.append({
 122                         "blocker"    : domain,
 123                         "blocked"    : blocked,
 124                         "reason"     : None,
 125                         "block_level": block_level,
 126                     })
 127
 128     elif "quarantined_instances" in data:
 129         logger.debug("Found 'quarantined_instances' in JSON response: domain='%s'", domain)
 130         found = True
 131         block_level = "quarantined"
 132
 133         for blocked in data["quarantined_instances"]:
 134             logger.debug("blocked='%s' - BEFORE!", blocked)
 135             blocked = tidyup.domain(blocked)
 136             logger.debug("blocked='%s' - AFTER!", blocked)
 137
 138             if blocked == "":
 139                 logger.warning("blocked is empty after tidyup.domain(): domain='%s',block_level='%s'", domain, block_level)
 140                 continue
 141             elif not utils.is_domain_wanted(blocked):
 142                 logger.debug("blocked='%s' is not wanted - SKIPPED!", blocked)
 143                 continue
 144
 145             logger.debug("Invoking utils.deobfuscate_domain(%s, %s) ...", blocked, domain)
 146             blocked = utils.deobfuscate_domain(blocked, domain)
 147
 148             logger.debug("blocked='%s' - DEOBFUSCATED!", blocked)
 149             if not utils.is_domain_wanted(blocked):
 150                 logger.debug("blocked='%s' is not wanted - SKIPPED!", blocked)
 151                 continue
 152
 153             logger.debug("Appending blocker='%s',blocked='%s',block_level='%s' ...", domain, blocked, block_level)
 154             blockdict.append({
 155                 "blocker"    : domain,
 156                 "blocked"    : blocked,
 157                 "reason"     : None,
 158                 "block_level": block_level,
 159             })
 160
 161     else:
 162         logger.warning("Cannot find 'mrf_simple' or 'quarantined_instances' in JSON reply: domain='%s'", domain)
 163
 164     logger.debug("Invoking commit() ...")
 165     database.connection.commit()
 166
 167     # Reasons
 168     if "mrf_simple_info" in data:
 169         logger.debug("Found mrf_simple_info in API response: domain='%s'", domain)
 170         found = True
 171         for block_level, info in (
 172             {
 173                 **data["mrf_simple_info"],
 174                 **(data["quarantined_instances_info"] if "quarantined_instances_info" in data else {})
 175             }
 176         ).items():
 177             logger.debug("block_level='%s', info.items()=%d", block_level, len(info.items()))
 178             block_level = tidyup.domain(block_level)
 179             logger.debug("block_level='%s' - AFTER!", block_level)
 180
 181             if block_level == "":
 182                 logger.warning("block_level is now empty!")
 183                 continue
 184             elif block_level == "accept":
 185                 logger.debug("domain='%s': Skipping block_level='%s' ...", domain, block_level)
 186                 continue
 187
 188             block_level = utils.alias_block_level(block_level)
 189
 190             logger.debug("Checking %d entries from domain='%s',block_level='%s' ...", len(info.items()), domain, block_level)
 191             for blocked, reason in info.items():
 192                 logger.debug("blocked='%s',reason[%s]='%s' - BEFORE!", blocked, type(reason), reason)
 193                 blocked = tidyup.domain(blocked)
 194                 logger.debug("blocked='%s' - AFTER!", blocked)
 195
 196                 if isinstance(reason, str):
 197                     logger.debug("reason[] is a string")
 198                     reason = tidyup.reason(reason)
 199                 elif isinstance(reason, dict) and "reason" in reason:
 200                     logger.debug("reason[] is a dict")
 201                     reason = tidyup.reason(reason["reason"]) if isinstance(reason["reason"], str) else None
 202                 elif reason is not None:
 203                     raise ValueError(f"Cannot handle reason[]='{type(reason)}'")
 204
 205                 logger.debug("blocked='%s',reason='%s' - AFTER!", blocked, reason)
 206
 207                 if blocked == "":
 208                     logger.warning("blocked is empty after tidyup.domain(): domain='%s',block_level='%s'", domain, block_level)
 209                     continue
 210                 elif not utils.is_domain_wanted(blocked):
 211                     logger.debug("blocked='%s' is not wanted - SKIPPED!", blocked)
 212                     continue
 213
 214                 logger.debug("Invoking utils.deobfuscate_domain(%s, %s) ...", blocked, domain)
 215                 blocked = utils.deobfuscate_domain(blocked, domain)
 216                 logger.debug("blocked='%s' - DEOBFUSCATED!", blocked)
 217
 218                 logger.debug("Checking %d blockdict records ...", len(blockdict))
 219                 for block in blockdict:
 220                     logger.debug("block[blocked]='%s',blocked='%s'", block["blocked"], blocked)
 221                     if block["blocked"] == blocked:
 222                         logger.debug("Updating reason='%s' for blocker='%s'", reason, block["blocked"])
 223                         block["reason"] = reason
 224
 225     elif "quarantined_instances_info" in data and "quarantined_instances" in data["quarantined_instances_info"]:
 226         logger.debug("Found 'quarantined_instances_info' in JSON response: domain='%s'", domain)
 227         found = True
 228         block_level = "quarantined"
 229
 230         #print(data["quarantined_instances_info"])
 231         rows = data["quarantined_instances_info"]["quarantined_instances"]
 232         for blocked in rows:
 233             logger.debug("blocked='%s' - BEFORE!", blocked)
 234             blocked = tidyup.domain(blocked)
 235             logger.debug("blocked='%s' - AFTER!", blocked)
 236
 237             if blocked not in rows or "reason" not in rows[blocked]:
 238                 logger.warning("Cannot find blocked='%s' in rows()=%d,domain='%s' - BREAK!", blocked, len(rows), domain)
 239                 break
 240
 241             reason = rows[blocked]["reason"]
 242             logger.debug("reason='%s'", reason)
 243
 244             if blocked == "":
 245                 logger.warning("blocked is empty after tidyup.domain(): domain='%s',block_level='%s'", domain, block_level)
 246                 continue
 247             elif not utils.is_domain_wanted(blocked):
 248                 logger.debug("blocked='%s' is not wanted - SKIPPED!", blocked)
 249                 continue
 250
 251             logger.debug("Invoking utils.deobfuscate_domain(%s, %s) ...", blocked, domain)
 252             blocked = utils.deobfuscate_domain(blocked, domain)
 253
 254             logger.debug("blocked='%s' - DEOBFUSCATED!", blocked)
 255             if not utils.is_domain_wanted(blocked):
 256                 logger.debug("blocked='%s' is not wanted - SKIPPED!", blocked)
 257                 continue
 258
 259             logger.debug("Checking %d blockdict records ...", len(blockdict))
 260             for block in blockdict:
 261                 logger.debug("block[blocked]='%s',blocked='%s'", block["blocked"], blocked)
 262                 if block["blocked"] == blocked:
 263                     logger.debug("Updating reason='%s' for blocker='%s'", reason, block["blocked"])
 264                     block["reason"] = reason
 265     else:
 266         logger.warning("Cannot find 'mrf_simple_info' or 'quarantined_instances_info' in JSON reply: domain='%s'", domain)
 267
 268     if not found:
 269         logger.debug("Did not find any useable JSON elements, domain='%s', continuing with /about page ...", domain)
 270         blocklist = fetch_blocks_from_about(domain)
 271
 272         logger.debug("blocklist()=%d", len(blocklist))
 273         if len(blocklist) > 0:
 274             logger.info("Checking %d record(s) ...", len(blocklist))
 275             for block_level in blocklist:
 276                 logger.debug("block_level='%s'", block_level)
 277                 rows = blocklist[block_level]
 278
 279                 logger.debug("rows[%s]()=%d'", type(rows), len(rows))
 280                 for record in rows:
 281                     logger.debug("record[]='%s'", type(record))
 282                     blocked = tidyup.domain(record["blocked"])
 283                     reason  = tidyup.reason(record["reason"])
 284                     logger.debug("blocked='%s',reason='%s' - AFTER!", blocked, reason)
 285
 286                     if blocked == "":
 287                         logger.warning("blocked is empty after tidyup.domain(): domain='%s',block_level='%s'", domain, block_level)
 288                         continue
 289                     elif not utils.is_domain_wanted(blocked):
 290                         logger.warning("blocked='%s' is not wanted - SKIPPED!", blocked)
 291                         continue
 292
 293                     logger.debug("Invoking utils.deobfuscate_domain(%s, %s) ...", blocked, domain)
 294                     blocked = utils.deobfuscate_domain(blocked, domain)
 295                     logger.debug("blocked='%s' - DEOBFUSCATED!", blocked)
 296
 297                     logger.debug("Appending blocker='%s',blocked='%s',reason='%s',block_level='%s' ...",domain, blocked, reason, block_level)
 298                     blockdict.append({
 299                         "blocker"    : domain,
 300                         "blocked"    : blocked,
 301                         "reason"     : reason,
 302                         "block_level": block_level,
 303                     })
 304
 305     logger.debug("blockdict()=%d - EXIT!", len(blockdict))
 306     return blockdict
 307
 308 def fetch_blocks_from_about(domain: str) -> dict:
 309     logger.debug("domain='%s' - CALLED!", domain)
 310     domain_helper.raise_on(domain)
 311
 312     logger.debug("Fetching mastodon blocks from domain='%s'", domain)
 313     doc = None
 314     for path in ["/instance/about/index.html"]:
 315         try:
 316             # Resetting doc type
 317             doc = None
 318
 319             logger.debug("Fetching path='%s' from domain='%s' ...", path, domain)
 320             response = network.fetch_response(
 321                 domain,
 322                 path,
 323                 network.web_headers,
 324                 (config.get("connection_timeout"), config.get("read_timeout"))
 325             )
 326
 327             logger.debug("response.ok='%s',response.status_code=%d,response.text()=%d", response.ok, response.status_code, len(response.text))
 328             if not response.ok or response.text.strip() == "":
 329                 logger.warning("path='%s' does not exist on domain='%s' - SKIPPED!", path, domain)
 330                 continue
 331
 332             logger.debug("Parsing response.text()=%d Bytes ...", len(response.text))
 333             doc = bs4.BeautifulSoup(
 334                 response.text,
 335                 "html.parser",
 336             )
 337
 338             logger.debug("doc[]='%s'", type(doc))
 339             if doc.find("h2") is not None:
 340                 logger.debug("Found 'h2' header in path='%s' - BREAK!", path)
 341                 break
 342
 343         except network.exceptions as exception:
 344             logger.warning("Cannot fetch from domain='%s',exception[%s]='%s'", domain, type(exception), str(exception))
 345             instances.set_last_error(domain, exception)
 346             break
 347
 348     blocklist = {
 349         "Suspended servers": [],
 350         "Filtered media"   : [],
 351         "Limited servers"  : [],
 352         "Silenced servers" : [],
 353     }
 354
 355     logger.debug("doc[]='%s'", type(doc))
 356     if doc is None:
 357         logger.warning("Cannot fetch any /about pages for domain='%s' - EXIT!", domain)
 358         return list()
 359
 360     for header in doc.find_all("h2"):
 361         logger.debug("header[%s]='%s'", type(header), header)
 362         header_text = tidyup.reason(header.text)
 363
 364         logger.debug("header_text='%s' - BEFORE!", header_text)
 365         if header_text in language_mapping:
 366             logger.debug("header_text='%s' - FOUND!", header_text)
 367             header_text = language_mapping[header_text]
 368         else:
 369             logger.warning("header_text='%s' not found in language mapping table", header_text)
 370
 371         logger.debug("header_text='%s - AFTER!'", header_text)
 372         if header_text in blocklist or header_text.lower() in blocklist:
 373             # replaced find_next_siblings with find_all_next to account for instances that e.g. hide lists in dropdown menu
 374             logger.debug("Found header_text='%s', importing domain blocks ...", header_text)
 375             for line in header.find_next("table").find_all("tr")[1:]:
 376                 logger.debug("line[]='%s'", type(line))
 377                 blocklist[header_text].append({
 378                     "blocked": tidyup.domain(line.find_all("td")[0].text),
 379                     "reason" : tidyup.reason(line.find_all("td")[1].text),
 380                 })
 381         else:
 382             logger.warning("header_text='%s' not found in blocklist()=%d", header_text, len(blocklist))
 383
 384     logger.debug("Returning blocklist for domain='%s' - EXIT!", domain)
 385     return {
 386         "reject"        : blocklist["Suspended servers"],
 387         "media_removal" : blocklist["Filtered media"],
 388         "followers_only": blocklist["Limited servers"] + blocklist["Silenced servers"],
 389     }