fba/networks/mastodon.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 inspect
  18
  19 import bs4
  20 import validators
  21
  22 from fba import blacklist
  23 from fba import blocks
  24 from fba import config
  25 from fba import fba
  26 from fba import instances
  27 from fba import network
  28 from fba.helpers import tidyup
  29
  30 language_mapping = {
  31     # English -> English
  32     "Silenced instances"            : "Silenced servers",
  33     "Suspended instances"           : "Suspended servers",
  34     "Limited instances"             : "Limited servers",
  35     "Filtered media"                : "Filtered media",
  36     # Mappuing German -> English
  37     "Gesperrte Server"              : "Suspended servers",
  38     "Gefilterte Medien"             : "Filtered media",
  39     "Stummgeschaltete Server"       : "Silenced servers",
  40     # Japanese -> English
  41     "停止済みのサーバー"            : "Suspended servers",
  42     "制限中のサーバー"              : "Limited servers",
  43     "メディアを拒否しているサーバー": "Filtered media",
  44     "サイレンス済みのサーバー"      : "Silenced servers",
  45     # ??? -> English
  46     "שרתים מושעים"                  : "Suspended servers",
  47     "מדיה מסוננת"                   : "Filtered media",
  48     "שרתים מוגבלים"                 : "Silenced servers",
  49     # French -> English
  50     "Serveurs suspendus"            : "Suspended servers",
  51     "Médias filtrés"                : "Filtered media",
  52     "Serveurs limités"              : "Limited servers",
  53     "Serveurs modérés"              : "Limited servers",
  54 }
  55
  56 def fetch_blocks_from_about(domain: str) -> dict:
  57     # DEBUG: print(f"DEBUG: domain='{domain}' - CALLED!")
  58     if not isinstance(domain, str):
  59         raise ValueError(f"Parameter domain[]={type(domain)} is not 'str'")
  60     elif domain == "":
  61         raise ValueError("Parameter 'domain' is empty")
  62
  63     # DEBUG: print("DEBUG: Fetching mastodon blocks from domain:", domain)
  64     blocklist = {
  65         "Suspended servers": [],
  66         "Filtered media"   : [],
  67         "Limited servers"  : [],
  68         "Silenced servers" : [],
  69     }
  70
  71     try:
  72         doc = bs4.BeautifulSoup(
  73             network.fetch_response(
  74                 domain,
  75                 "/about/more",
  76                 network.web_headers,
  77                 (config.get("connection_timeout"), config.get("read_timeout"))
  78             ).text,
  79             "html.parser",
  80         )
  81     except BaseException as exception:
  82         print("ERROR: Cannot fetch from domain:", domain, exception)
  83         instances.update_last_error(domain, exception)
  84         return {}
  85
  86     for header in doc.find_all("h3"):
  87         header_text = tidyup.reason(header.text)
  88
  89         # DEBUG: print(f"DEBUG: header_text='{header_text}'")
  90         if header_text in language_mapping:
  91             # DEBUG: print(f"DEBUG: header_text='{header_text}'")
  92             header_text = language_mapping[header_text]
  93         else:
  94             print(f"WARNING: header_text='{header_text}' not found in language mapping table")
  95
  96         if header_text in blocklist or header_text.lower() in blocklist:
  97             # replaced find_next_siblings with find_all_next to account for instances that e.g. hide lists in dropdown menu
  98             for line in header.find_all_next("table")[0].find_all("tr")[1:]:
  99                 blocklist[header_text].append(
 100                     {
 101                         "domain": tidyup.domain(line.find("span").text),
 102                         "hash"  : tidyup.domain(line.find("span")["title"][9:]),
 103                         "reason": tidyup.reason(line.find_all("td")[1].text),
 104                     }
 105                 )
 106         else:
 107             print(f"WARNING: header_text='{header_text}' not found in blocklist()={len(blocklist)}")
 108
 109     # DEBUG: print("DEBUG: Returning blocklist for domain:", domain)
 110     return {
 111         "reject"        : blocklist["Suspended servers"],
 112         "media_removal" : blocklist["Filtered media"],
 113         "followers_only": blocklist["Limited servers"] + blocklist["Silenced servers"],
 114     }
 115
 116 def fetch_blocks(domain: str, origin: str, nodeinfo_url: str):
 117     # DEBUG: print(f"DEBUG: domain='{domain}',origin='{origin}',nodeinfo_url='{nodeinfo_url}' - CALLED!")
 118     if not isinstance(domain, str):
 119         raise ValueError(f"Parameter domain[]={type(domain)} is not 'str'")
 120     elif domain == "":
 121         raise ValueError("Parameter 'domain' is empty")
 122     elif not isinstance(origin, str) and origin is not None:
 123         raise ValueError(f"Parameter origin[]={type(origin)} is not 'str'")
 124     elif origin == "":
 125         raise ValueError("Parameter 'origin' is empty")
 126     elif not isinstance(nodeinfo_url, str):
 127         raise ValueError(f"Parameter nodeinfo_url[]={type(nodeinfo_url)} is not 'str'")
 128     elif nodeinfo_url == "":
 129         raise ValueError("Parameter 'nodeinfo_url' is empty")
 130
 131     try:
 132         # json endpoint for newer mastodongs
 133         blockdict = list()
 134         try:
 135             rows = {
 136                 "reject"        : [],
 137                 "media_removal" : [],
 138                 "followers_only": [],
 139                 "report_removal": [],
 140             }
 141
 142             # DEBUG: print("DEBUG: Querying API domain_blocks:", domain)
 143             response = network.fetch_response(
 144                 domain,
 145                 "/api/v1/instance/domain_blocks",
 146                 network.api_headers,
 147                 (config.get("connection_timeout"), config.get("read_timeout"))
 148             )
 149
 150             # DEBUG: print(f"DEBUG: response[]='{type(response)}'")
 151             blocklist = network.json_from_response(response)
 152
 153             print(f"INFO: Checking {len(blocklist)} entries from domain='{domain}',software='mastodon' ...")
 154             for block in blocklist:
 155                 entry = {
 156                     'domain': block['domain'],
 157                     'hash'  : block['digest'],
 158                     'reason': block['comment']
 159                 }
 160
 161                 # DEBUG: print("DEBUG: severity,domain,hash,comment:", block['severity'], block['domain'], block['digest'], block['comment'])
 162                 if block['severity'] == 'suspend':
 163                     # DEBUG: print(f"DEBUG: Adding entry='{entry}' with severity='{block['severity']}' ...")
 164                     rows['reject'].append(entry)
 165                 elif block['severity'] == 'silence':
 166                     # DEBUG: print(f"DEBUG: Adding entry='{entry}' with severity='{block['severity']}' ...")
 167                     rows['followers_only'].append(entry)
 168                 elif block['severity'] == 'reject_media':
 169                     # DEBUG: print(f"DEBUG: Adding entry='{entry}' with severity='{block['severity']}' ...")
 170                     rows['media_removal'].append(entry)
 171                 elif block['severity'] == 'reject_reports':
 172                     # DEBUG: print(f"DEBUG: Adding entry='{entry}' with severity='{block['severity']}' ...")
 173                     rows['report_removal'].append(entry)
 174                 else:
 175                     print("WARNING: Unknown severity:", block['severity'], block['domain'])
 176
 177         except BaseException as exception:
 178             # DEBUG: print(f"DEBUG: Failed, trying mastodon-specific fetches: domain='{domain}',exception[{type(exception)}]={str(exception)}")
 179             rows = fetch_blocks_from_about(domain)
 180
 181         print(f"INFO: Checking {len(rows.items())} entries from domain='{domain}',software='mastodon' ...")
 182         for block_level, blocklist in rows.items():
 183             # DEBUG: print("DEBUG: domain,block_level,blocklist():", domain, block_level, len(blocklist))
 184             block_level = tidyup.domain(block_level)
 185
 186             # DEBUG: print("DEBUG: AFTER-block_level:", block_level)
 187             if block_level == "":
 188                 print("WARNING: block_level is empty, domain:", domain)
 189                 continue
 190
 191             # DEBUG: print(f"DEBUG: Checking {len(blocklist)} entries from domain='{domain}',software='mastodon',block_level='{block_level}' ...")
 192             for block in blocklist:
 193                 # DEBUG: print(f"DEBUG: block[]='{type(block)}'")
 194                 blocked, blocked_hash, reason = block.values()
 195                 # DEBUG: print(f"DEBUG: blocked='{blocked}',blocked_hash='{blocked_hash}',reason='{reason}':")
 196                 blocked = tidyup.domain(blocked)
 197                 reason  = tidyup.reason(reason) if reason is not None and reason != "" else None
 198                 # DEBUG: print(f"DEBUG: blocked='{blocked}',reason='{reason}' - AFTER!")
 199
 200                 if blocked == "":
 201                     print("WARNING: blocked is empty:", domain)
 202                     continue
 203                 elif blacklist.is_blacklisted(blocked):
 204                     # DEBUG: print(f"DEBUG: blocked='{blocked}' is blacklisted - skipping!")
 205                     continue
 206                 elif blocked.count("*") > 0:
 207                     # Doing the hash search for instance names as well to tidy up DB
 208                     fba.cursor.execute(
 209                         "SELECT domain, origin, nodeinfo_url FROM instances WHERE hash = ? LIMIT 1", [blocked_hash]
 210                     )
 211                     searchres = fba.cursor.fetchone()
 212
 213                     if searchres is None:
 214                         print(f"WARNING: Cannot deobsfucate blocked='{blocked}',blocked_hash='{blocked_hash}' - SKIPPED!")
 215                         continue
 216
 217                     # DEBUG: print("DEBUG: Updating domain: ", searchres[0])
 218                     blocked = searchres[0]
 219                     origin = searchres[1]
 220                     nodeinfo_url = searchres[2]
 221
 222                     # DEBUG: print("DEBUG: Looking up instance by domain:", blocked)
 223                     if not validators.domain(blocked):
 224                         print(f"WARNING: blocked='{blocked}',software='mastodon' is not a valid domain name - skipped!")
 225                         continue
 226                     elif not instances.is_registered(blocked):
 227                         # DEBUG: print(f"DEBUG: Domain blocked='{blocked}' wasn't found, adding ..., domain='{domain}',origin='{origin}',nodeinfo_url='{nodeinfo_url}'")
 228                         instances.add(blocked, domain, inspect.currentframe().f_code.co_name, nodeinfo_url)
 229                 elif not validators.domain(blocked):
 230                     print(f"WARNING: blocked='{blocked}',software='mastodon' is not a valid domain name - skipped!")
 231                     continue
 232
 233                 # DEBUG: print("DEBUG: Looking up instance by domain:", blocked)
 234                 if not validators.domain(blocked):
 235                     print(f"WARNING: blocked='{blocked}',software='mastodon' is not a valid domain name - skipped!")
 236                     continue
 237                 elif not instances.is_registered(blocked):
 238                     # DEBUG: print("DEBUG: Hash wasn't found, adding:", blocked, domain)
 239                     instances.add(blocked, domain, inspect.currentframe().f_code.co_name, nodeinfo_url)
 240
 241                 blocking = blocked if blocked.count("*") <= 1 else blocked_hash
 242                 # DEBUG: print(f"DEBUG: blocking='{blocking}',blocked='{blocked}',blocked_hash='{blocked_hash}'")
 243
 244                 if not blocks.is_instance_blocked(domain, blocked, block_level):
 245                     # DEBUG: print("DEBUG: Blocking:", domain, blocked, block_level)
 246                     blocks.add_instance(domain, blocking, reason, block_level)
 247
 248                     if block_level == "reject":
 249                         blockdict.append({
 250                             "blocked": blocked,
 251                             "reason" : reason
 252                         })
 253                 else:
 254                     # DEBUG: print(f"DEBUG: Updating block last seen and reason for domain='{domain}',blocking='{blocking}' ...")
 255                     blocks.update_last_seen(domain, blocking, block_level)
 256                     blocks.update_reason(reason, domain, blocking, block_level)
 257
 258         # DEBUG: print("DEBUG: Committing changes ...")
 259         fba.connection.commit()
 260     except BaseException as exception:
 261         print(f"ERROR: domain='{domain}',software='mastodon',exception[{type(exception)}]:'{str(exception)}'")
 262
 263     # DEBUG: print("DEBUG: EXIT!")