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             elif block_level == "suspend":
  96                 logger.debug("domain='%s', mapping 'suspend' to 'suspended'", domain)
  97                 block_level = "suspended"
  98             elif block_level == "silence":
  99                 logger.debug("domain='%s', mapping 'silence' to 'silenced'", domain)
 100                 block_level = "silenced"
 101
 102             logger.debug("Checking %d entries from domain='%s',block_level='%s' ...", len(blocklist), domain, block_level)
 103             if len(blocklist) > 0:
 104                 for blocked in blocklist:
 105                     logger.debug("blocked='%s' - BEFORE!", blocked)
 106                     blocked = tidyup.domain(blocked)
 107                     logger.debug("blocked='%s' - AFTER!", blocked)
 108
 109                     if blocked == "":
 110                         logger.warning("blocked is empty after tidyup.domain(): domain='%s',block_level='%s'", domain, block_level)
 111                         continue
 112                     elif not utils.is_domain_wanted(blocked):
 113                         logger.debug("blocked='%s' is not wanted - SKIPPED!", blocked)
 114                         continue
 115
 116                     logger.debug("Invoking utils.deobfuscate_domain(%s, %s) ...", blocked, domain)
 117                     blocked = utils.deobfuscate_domain(blocked, domain)
 118
 119                     logger.debug("blocked='%s' - DEOBFUSCATED!", blocked)
 120                     if not utils.is_domain_wanted(blocked):
 121                         logger.debug("blocked='%s' is not wanted - SKIPPED!", blocked)
 122                         continue
 123
 124                     logger.debug("Appending blocker='%s',blocked='%s',block_level='%s' ...", domain, blocked, block_level)
 125                     blockdict.append({
 126                         "blocker"    : domain,
 127                         "blocked"    : blocked,
 128                         "reason"     : None,
 129                         "block_level": block_level,
 130                     })
 131
 132     elif "quarantined_instances" in data:
 133         logger.debug("Found 'quarantined_instances' in JSON response: domain='%s'", domain)
 134         found = True
 135         block_level = "quarantined"
 136
 137         for blocked in data["quarantined_instances"]:
 138             logger.debug("blocked='%s' - BEFORE!", blocked)
 139             blocked = tidyup.domain(blocked)
 140             logger.debug("blocked='%s' - AFTER!", blocked)
 141
 142             if blocked == "":
 143                 logger.warning("blocked is empty after tidyup.domain(): domain='%s',block_level='%s'", domain, block_level)
 144                 continue
 145             elif not utils.is_domain_wanted(blocked):
 146                 logger.debug("blocked='%s' is not wanted - SKIPPED!", blocked)
 147                 continue
 148
 149             logger.debug("Invoking utils.deobfuscate_domain(%s, %s) ...", blocked, domain)
 150             blocked = utils.deobfuscate_domain(blocked, domain)
 151
 152             logger.debug("blocked='%s' - DEOBFUSCATED!", blocked)
 153             if not utils.is_domain_wanted(blocked):
 154                 logger.debug("blocked='%s' is not wanted - SKIPPED!", blocked)
 155                 continue
 156
 157             logger.debug("Appending blocker='%s',blocked='%s',block_level='%s' ...", domain, blocked, block_level)
 158             blockdict.append({
 159                 "blocker"    : domain,
 160                 "blocked"    : blocked,
 161                 "reason"     : None,
 162                 "block_level": block_level,
 163             })
 164
 165     else:
 166         logger.warning("Cannot find 'mrf_simple' or 'quarantined_instances' in JSON reply: domain='%s'", domain)
 167
 168     logger.debug("Invoking commit() ...")
 169     database.connection.commit()
 170
 171     # Reasons
 172     if "mrf_simple_info" in data:
 173         logger.debug("Found mrf_simple_info in API response: domain='%s'", domain)
 174         found = True
 175         for block_level, info in (
 176             {
 177                 **data["mrf_simple_info"],
 178                 **(data["quarantined_instances_info"] if "quarantined_instances_info" in data else {})
 179             }
 180         ).items():
 181             logger.debug("block_level='%s', info.items()=%d", block_level, len(info.items()))
 182             block_level = tidyup.domain(block_level)
 183             logger.debug("block_level='%s' - AFTER!", block_level)
 184
 185             if block_level == "":
 186                 logger.warning("block_level is now empty!")
 187                 continue
 188             elif block_level == "accept":
 189                 logger.debug("domain='%s' skipping block_level='accept'", domain)
 190                 continue
 191             elif block_level == "suspend":
 192                 logger.debug("domain='%s', mapping 'suspend' to 'suspended'", domain)
 193                 block_level = "suspended"
 194             elif block_level == "silence":
 195                 logger.debug("domain='%s', mapping 'silence' to 'silenced'", domain)
 196                 block_level = "silenced"
 197
 198             logger.debug("Checking %d entries from domain='%s',block_level='%s' ...", len(info.items()), domain, block_level)
 199             for blocked, reason in info.items():
 200                 logger.debug("blocked='%s',reason[%s]='%s' - BEFORE!", blocked, type(reason), reason)
 201                 blocked = tidyup.domain(blocked)
 202                 logger.debug("blocked='%s' - AFTER!", blocked)
 203
 204                 if isinstance(reason, str):
 205                     logger.debug("reason[] is a string")
 206                     reason = tidyup.reason(reason)
 207                 elif isinstance(reason, dict) and "reason" in reason:
 208                     logger.debug("reason[] is a dict")
 209                     reason = tidyup.reason(reason["reason"])
 210                 elif reason is not None:
 211                     raise ValueError(f"Cannot handle reason[]='{type(reason)}'")
 212
 213                 logger.debug("blocked='%s',reason='%s' - AFTER!", blocked, reason)
 214
 215                 if blocked == "":
 216                     logger.warning("blocked is empty after tidyup.domain(): domain='%s',block_level='%s'", domain, block_level)
 217                     continue
 218                 elif not utils.is_domain_wanted(blocked):
 219                     logger.debug("blocked='%s' is not wanted - SKIPPED!", blocked)
 220                     continue
 221
 222                 logger.debug("Invoking utils.deobfuscate_domain(%s, %s) ...", blocked, domain)
 223                 blocked = utils.deobfuscate_domain(blocked, domain)
 224                 logger.debug("blocked='%s' - DEOBFUSCATED!", blocked)
 225
 226                 logger.debug("Checking %d blockdict records ...", len(blockdict))
 227                 for block in blockdict:
 228                     logger.debug("block[blocked]='%s',blocked='%s'", block['blocked'], blocked)
 229                     if block['blocked'] == blocked:
 230                         logger.debug("Updating reason='%s' for blocker='%s'", reason, block['blocked'])
 231                         block['reason'] = reason
 232
 233     elif "quarantined_instances_info" in data and "quarantined_instances" in data["quarantined_instances_info"]:
 234         logger.debug("Found 'quarantined_instances_info' in JSON response: domain='%s'", domain)
 235         found = True
 236         block_level = "quarantined"
 237
 238         #print(data["quarantined_instances_info"])
 239         rows = data["quarantined_instances_info"]["quarantined_instances"]
 240         for blocked in rows:
 241             logger.debug("blocked='%s' - BEFORE!", blocked)
 242             blocked = tidyup.domain(blocked)
 243             logger.debug("blocked='%s' - AFTER!", blocked)
 244
 245             if blocked not in rows or "reason" not in rows[blocked]:
 246                 logger.warning("Cannot find blocked='%s' in rows()=%d,domain='%s' - BREAK!", blocked, len(rows), domain)
 247                 break
 248
 249             reason = rows[blocked]["reason"]
 250             logger.debug("reason='%s'", reason)
 251
 252             if blocked == "":
 253                 logger.warning("blocked is empty after tidyup.domain(): domain='%s',block_level='%s'", domain, block_level)
 254                 continue
 255
 256             logger.debug("Invoking utils.deobfuscate_domain(%s, %s) ...", blocked, domain)
 257             blocked = utils.deobfuscate_domain(blocked, domain)
 258
 259             logger.debug("blocked='%s' - DEOBFUSCATED!", blocked)
 260             if not utils.is_domain_wanted(blocked):
 261                 logger.debug("blocked='%s' is not wanted - SKIPPED!", blocked)
 262                 continue
 263
 264             logger.debug("Checking %d blockdict records ...", len(blockdict))
 265             for block in blockdict:
 266                 logger.debug("block[blocked]='%s',blocked='%s'", block['blocked'], blocked)
 267                 if block['blocked'] == blocked:
 268                     logger.debug("Updating reason='%s' for blocker='%s'", reason, block['blocked'])
 269                     block['reason'] = reason
 270     else:
 271         logger.warning("Cannot find 'mrf_simple_info' or 'quarantined_instances_info' in JSON reply: domain='%s'", domain)
 272
 273     if not found:
 274         logger.debug("Did not find any useable JSON elements, domain='%s', continuing with /about page ...", domain)
 275         blocklist = fetch_blocks_from_about(domain)
 276
 277         logger.debug("blocklist()=%d", len(blocklist))
 278         if len(blocklist) > 0:
 279             logger.info("Checking %d record(s) ...", len(blocklist))
 280             for block_level in blocklist:
 281                 logger.debug("block_level='%s'", block_level)
 282                 rows = blocklist[block_level]
 283
 284                 logger.debug("rows[%s]()=%d'", type(rows), len(rows))
 285                 for record in rows:
 286                     logger.debug("record[]='%s'", type(record))
 287                     blocked = tidyup.domain(record["blocked"])
 288                     reason  = tidyup.reason(record["reason"])
 289                     logger.debug("blocked='%s',reason='%s' - AFTER!", blocked, reason)
 290
 291                     if blocked == "":
 292                         logger.warning("blocked is empty after tidyup.domain(): domain='%s',block_level='%s'", domain, block_level)
 293                         continue
 294                     elif not utils.is_domain_wanted(blocked):
 295                         logger.warning("blocked='%s' is not wanted - SKIPPED!", blocked)
 296                         continue
 297
 298                     logger.debug("Invoking utils.deobfuscate_domain(%s, %s) ...", blocked, domain)
 299                     blocked = utils.deobfuscate_domain(blocked, domain)
 300                     logger.debug("blocked='%s' - DEOBFUSCATED!", blocked)
 301
 302                     logger.debug("Appending blocker='%s',blocked='%s',reason='%s',block_level='%s' ...",domain, blocked, reason, block_level)
 303                     blockdict.append({
 304                         "blocker"    : domain,
 305                         "blocked"    : blocked,
 306                         "reason"     : reason,
 307                         "block_level": block_level,
 308                     })
 309
 310     logger.debug("blockdict()=%d - EXIT!", len(blockdict))
 311     return blockdict
 312
 313 def fetch_blocks_from_about(domain: str) -> dict:
 314     logger.debug("domain='%s' - CALLED!", domain)
 315     domain_helper.raise_on(domain)
 316
 317     logger.debug("Fetching mastodon blocks from domain='%s'", domain)
 318     doc = None
 319     for path in ["/instance/about/index.html"]:
 320         try:
 321             # Resetting doc type
 322             doc = None
 323
 324             logger.debug("Fetching path='%s' from domain='%s' ...", path, domain)
 325             response = network.fetch_response(
 326                 domain,
 327                 path,
 328                 network.web_headers,
 329                 (config.get("connection_timeout"), config.get("read_timeout"))
 330             )
 331
 332             logger.debug("response.ok='%s',response.status_code='%d',response.text()=%d", response.ok, response.status_code, len(response.text))
 333             if not response.ok or response.text.strip() == "":
 334                 logger.warning("path='%s' does not exist on domain='%s' - SKIPPED!", path, domain)
 335                 continue
 336
 337             logger.debug("Parsing response.text()=%d Bytes ...", len(response.text))
 338             doc = bs4.BeautifulSoup(
 339                 response.text,
 340                 "html.parser",
 341             )
 342
 343             logger.debug("doc[]='%s'", type(doc))
 344             if doc.find("h2") is not None:
 345                 logger.debug("Found 'h2' header in path='%s' - BREAK!", path)
 346                 break
 347
 348         except network.exceptions as exception:
 349             logger.warning("Cannot fetch from domain='%s',exception[%s]='%s'", domain, type(exception), str(exception))
 350             instances.set_last_error(domain, exception)
 351             break
 352
 353     blocklist = {
 354         "Suspended servers": [],
 355         "Filtered media"   : [],
 356         "Limited servers"  : [],
 357         "Silenced servers" : [],
 358     }
 359
 360     logger.debug("doc[]='%s'", type(doc))
 361     if doc is None:
 362         logger.warning("Cannot fetch any /about pages for domain='%s' - EXIT!", domain)
 363         return list()
 364
 365     for header in doc.find_all("h2"):
 366         header_text = tidyup.reason(header.text)
 367
 368         logger.debug("header_text='%s' - BEFORE!", header_text)
 369         if header_text in language_mapping:
 370             logger.debug("header_text='%s' - FOUND!", header_text)
 371             header_text = language_mapping[header_text]
 372         else:
 373             logger.warning("header_text='%s' not found in language mapping table", header_text)
 374
 375         logger.debug("header_text='%s - AFTER!'", header_text)
 376         if header_text in blocklist or header_text.lower() in blocklist:
 377             # replaced find_next_siblings with find_all_next to account for instances that e.g. hide lists in dropdown menu
 378             logger.debug("Found header_text='%s', importing domain blocks ...", header_text)
 379             for line in header.find_next("table").find_all("tr")[1:]:
 380                 logger.debug("line[]='%s'", type(line))
 381                 blocklist[header_text].append({
 382                     "blocked": tidyup.domain(line.find_all("td")[0].text),
 383                     "reason" : tidyup.reason(line.find_all("td")[1].text),
 384                 })
 385         else:
 386             logger.warning("header_text='%s' not found in blocklist()=%d", header_text, len(blocklist))
 387
 388     logger.debug("Returning blocklist for domain='%s' - EXIT!", domain)
 389     return {
 390         "reject"        : blocklist["Suspended servers"],
 391         "media_removal" : blocklist["Filtered media"],
 392         "followers_only": blocklist["Limited servers"] + blocklist["Silenced servers"],
 393     }