1 # Fedi API Block - An aggregator for fetching blocking data from fediverse nodes
2 # Copyright (C) 2023 Free Software Foundation
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.
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.
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/>.
24 from fba import database
27 from fba.helpers import blacklist
28 from fba.helpers import cache
29 from fba.helpers import config
30 from fba.helpers import domain as domain_helper
31 from fba.helpers import tidyup
33 from fba.http import federation
34 from fba.http import network
36 from fba.models import error_log
38 logging.basicConfig(level=logging.INFO)
39 logger = logging.getLogger(__name__)
40 #logger.setLevel(logging.DEBUG)
42 # Found info from node, such as nodeinfo URL, detection mode that needs to be
43 # written to database. Both arrays must be filled at the same time or else
47 # NULL means all detection methods have failed (maybe still reachable instance)
48 "detection_mode" : {},
56 "obfuscated_blocks" : {},
57 # Last fetched instances
58 "last_instance_fetch": {},
63 # Last nodeinfo (fetched)
66 "last_status_code" : {},
68 "last_error_details" : {},
69 # Wether obfuscation has been used
70 "has_obfuscation" : {},
75 def _set_data(key: str, domain: str, value: any):
76 logger.debug("key='%s',domain='%s',value[]='%s' - CALLED!", key, domain, type(value))
77 domain_helper.raise_on(domain)
78 if not isinstance(key, str):
79 raise ValueError(f"Parameter key[]='{type(key)}' is not of type 'str'")
81 raise ValueError("Parameter 'key' is empty")
82 elif not key in _pending:
83 raise ValueError(f"key='{key}' not found in _pending")
84 elif not utils.is_primitive(value):
85 raise ValueError(f"value[]='{type(value)}' is not a primitive type")
88 _pending[key][domain] = value
92 def has_pending(domain: str) -> bool:
93 logger.debug("domain='%s' - CALLED!", domain)
94 domain_helper.raise_on(domain)
98 logger.debug("key='%s',domain='%s',_pending[key]()=%d", key, domain, len(_pending[key]))
99 if domain in _pending[key]:
103 logger.debug("has='%s' - EXIT!", has)
106 def update(domain: str):
107 logger.debug("domain='%s' - CALLED!", domain)
108 domain_helper.raise_on(domain)
109 if not has_pending(domain):
110 raise Exception(f"domain='{domain}' has no pending instance data, but function invoked")
111 elif not is_registered(domain):
112 raise Exception(f"domain='{domain}' cannot be updated while not being registered")
114 logger.debug("Updating instance data for domain='%s' ...", domain)
118 logger.debug("Checking key='%s',domain='%s'", key, domain)
119 if domain in _pending[key]:
120 logger.debug("Adding '%s' for key='%s' ...", _pending[key][domain], key)
121 fields.append(_pending[key][domain])
122 sql_string += f" {key} = ?,"
124 logger.debug("sql_string(%d)='%s'", len(sql_string), sql_string)
126 raise ValueError(f"No fields have been set, but function invoked, domain='{domain}'")
128 # Set last_updated to current timestamp
129 fields.append(time.time())
131 # For WHERE statement
132 logger.debug("Setting domain='%s' for WHERE statement ...", domain)
133 fields.append(domain)
135 logger.debug("sql_string='%s',fields()=%d", sql_string, len(fields))
136 sql_string = "UPDATE instances SET" + sql_string + " last_updated = ? WHERE domain = ? LIMIT 1"
138 logger.debug("Executing SQL: sql_string='%s',fields()=%d", sql_string, len(fields))
139 database.cursor.execute(sql_string, fields)
141 logger.debug("rowcount=%d", database.cursor.rowcount)
142 if database.cursor.rowcount == 0:
143 raise Exception(f"Did not update any rows: domain='{domain}',fields()={len(fields)}")
145 logger.debug("Invoking commit() ...")
146 database.connection.commit()
148 logger.debug("Deleting _pending for domain='%s'", domain)
150 logger.debug("domain='%s',key='%s'", domain, key)
151 if domain in _pending[key]:
152 logger.debug("Deleting key='%s',domain='%s' ...", key, domain)
153 del _pending[key][domain]
155 logger.debug("EXIT!")
157 def add(domain: str, origin: str, command: str, path: str = None, software: str = None):
158 logger.debug("domain='%s',origin='%s',command='%s',path='%s',software='%s' - CALLED!", domain, origin, command, path, software)
159 domain_helper.raise_on(domain)
161 if not isinstance(origin, str) and origin is not None:
162 raise ValueError(f"origin[]='{type(origin)}' is not of type 'str'")
164 raise ValueError("Parameter 'origin' is empty")
165 elif not isinstance(command, str):
166 raise ValueError(f"command[]='{type(command)}' is not of type 'str'")
168 raise ValueError("Parameter 'command' is empty")
169 elif not isinstance(path, str) and path is not None:
170 raise ValueError(f"path[]='{type(path)}' is not of type 'str'")
172 raise ValueError("Parameter 'path' is empty")
173 elif not isinstance(software, str) and software is not None:
174 raise ValueError(f"software[]='{type(software)}' is not of type 'str'")
176 raise ValueError("Parameter 'software' is empty")
177 elif origin is not None and not validators.domain(origin.split("/")[0]):
178 raise ValueError(f"Bad origin name='{origin}'")
179 elif blacklist.is_blacklisted(domain):
180 raise Exception(f"domain='{domain}' is blacklisted, but function invoked")
181 elif domain.find("/profile/") > 0 or domain.find("/users/") > 0 or (is_registered(domain.split("/")[0]) and domain.find("/c/") > 0):
182 raise Exception(f"domain='{domain}' is a single user")
183 elif domain.find("/tag/") > 0:
184 raise Exception(f"domain='{domain}' is a tag")
188 logger.debug("domain='%s',origin='%s',command='%s',path='%s'", domain, origin, command, path)
189 software = federation.determine_software(domain, path)
190 except network.exceptions as exception:
191 logger.warning("Exception '%s' during determining software type, domain='%s'", type(exception), domain)
192 set_last_error(domain, exception)
194 logger.debug("Determined software='%s'", software)
195 if software == "lemmy" and domain.find("/c/") > 0:
196 domain = domain.split("/c/")[0]
197 if is_registered(domain):
198 logger.warning("domain='%s' already registered after cutting off user part. - EXIT!", domain)
201 logger.info("Adding instance domain='%s',origin='%s',software='%s',command='%s'", domain, origin, software, command)
202 database.cursor.execute(
203 "INSERT INTO instances (domain, origin, command, hash, software, first_seen) VALUES (?, ?, ?, ?, ?, ?)",
208 utils.get_hash(domain),
214 logger.debug("Marking domain='%s' as registered.", domain)
215 cache.set_sub_key("is_registered", domain, True)
217 logger.debug("Checking if domain='%s' has pending updates ...", domain)
218 if has_pending(domain):
219 logger.debug("Flushing updates for domain='%s' ...", domain)
222 logger.debug("EXIT!")
224 def set_last_nodeinfo(domain: str):
225 logger.debug("domain='%s' - CALLED!", domain)
226 domain_helper.raise_on(domain)
228 logger.debug("Updating last_nodeinfo for domain='%s'", domain)
229 _set_data("last_nodeinfo", domain, time.time())
231 logger.debug("EXIT!")
233 def set_last_error(domain: str, error: dict):
234 logger.debug("domain='%s',error[]='%s' - CALLED!", domain, type(error))
235 domain_helper.raise_on(domain)
237 logger.debug("error[]='%s' - BEFORE!", type(error))
238 if isinstance(error, (BaseException, json.decoder.JSONDecodeError)):
239 error = f"error[{type(error)}]='{str(error)}'"
240 logger.debug("error[]='%s' - AFTER!", type(error))
242 if isinstance(error, str):
243 logger.debug("Setting last_error_details='%s' (str)", error)
244 _set_data("last_status_code" , domain, 999)
245 _set_data("last_error_details", domain, error if error != "" else None)
246 elif isinstance(error, requests.models.Response):
247 logger.debug("Setting last_error_details='%s' (Response)", error.reason)
248 _set_data("last_status_code" , domain, error.status_code)
249 _set_data("last_error_details", domain, error.reason if error.reason != "" else None)
250 elif not isinstance(error, dict):
251 raise KeyError(f"Cannot handle keys in error[{type(error)}]='{error}'")
252 elif "status_code" in error and "error_message" in error:
253 logger.debug("Setting last_error_details='%s' (error_message)", error['error_message'])
254 _set_data("last_status_code" , domain, error["status_code"])
255 _set_data("last_error_details", domain, error["error_message"] if error["error_message"] != "" else None)
256 elif "json" in error and "error" in error["json"]:
257 logger.debug("Setting last_error_details='%s' (json,error)", error["json"]["error"])
258 _set_data("last_status_code" , domain, error["status_code"])
259 _set_data("last_error_details", domain, error["json"]["error"] if error["json"]["error"] != "" else None)
261 logger.debug("Invoking error_log.add(domain='%s',error[]='%s'", domain, type(error))
262 error_log.add(domain, error)
264 logger.debug("EXIT!")
266 def set_success(domain: str):
267 logger.debug("domain='%s' - CALLED!", domain)
268 domain_helper.raise_on(domain)
270 # Set both to success
271 _set_data("last_status_code" , domain, 200)
272 _set_data("last_error_details", domain, None)
274 logger.debug("EXIT!")
276 def is_registered(domain: str, skip_raise = False) -> bool:
277 logger.debug("domain='%s',skip_raise='%s' - CALLED!", domain, skip_raise)
278 if not isinstance(skip_raise, bool):
279 raise ValueError(f"skip_raise[]='{type(skip_raise)}' is not type of 'bool'")
282 domain_helper.raise_on(domain)
284 logger.debug("domain='%s' - CALLED!", domain)
285 if not cache.key_exists("is_registered"):
286 logger.debug("Cache for 'is_registered' not initialized, fetching all rows ...")
287 database.cursor.execute("SELECT domain FROM instances")
290 cache.set_all("is_registered", database.cursor.fetchall(), True)
293 registered = cache.sub_key_exists("is_registered", domain)
295 logger.debug("registered='%s' - EXIT!", registered)
298 def is_recent(domain: str, column: str = "last_instance_fetch") -> bool:
299 logger.debug("domain='%s',column='%s' - CALLED!", domain, column)
300 domain_helper.raise_on(domain)
302 if not isinstance(column, str):
303 raise ValueError(f"Parameter column[]='{type(column)}' is not of type 'str'")
304 elif not column.startswith("last_"):
305 raise ValueError(f"Parameter column='{column}' is not expected")
306 elif not is_registered(domain):
307 logger.debug("domain='%s' is not registered, returning False - EXIT!", domain)
310 key = "recheck_instance"
311 if column == "last_blocked":
312 key = "recheck_block"
315 database.cursor.execute(f"SELECT {column} FROM instances WHERE domain = ? LIMIT 1", [domain])
318 row = database.cursor.fetchone()
320 fetched = float(row[column]) if row[column] is not None else 0.0
322 diff = (time.time() - fetched)
324 logger.debug("fetched[%s]='%s',key='%s',diff=%f", type(fetched), fetched, key, diff)
325 recently = bool(diff < config.get(key))
327 logger.debug("recently='%s' - EXIT!", recently)
330 def deobfuscate(char: str, domain: str, blocked_hash: str = None) -> tuple:
331 logger.debug("char='%s',domain='%s',blocked_hash='%s' - CALLED!", char, domain, blocked_hash)
333 if not isinstance(char, str):
334 raise ValueError(f"Parameter char[]='{type(char)}' is not of type 'str'")
336 raise ValueError("Parameter 'char' is empty")
337 elif not char in domain:
338 raise ValueError(f"char='{char}' not found in domain='{domain}' but function invoked")
339 elif not isinstance(domain, str):
340 raise ValueError(f"Parameter domain[]='{type(domain)}'")
342 raise ValueError("Parameter 'domain' is empty")
343 elif not isinstance(blocked_hash, str) and blocked_hash is not None:
344 raise ValueError(f"Parameter blocked_hash[]='{type(blocked_hash)}' is not of type 'str'")
349 logger.debug("blocked_hash[]='%s'", type(blocked_hash))
350 if isinstance(blocked_hash, str):
351 logger.debug("Looking up blocked_hash='%s',domain='%s' ...", blocked_hash, domain)
352 database.cursor.execute(
353 "SELECT domain, origin, nodeinfo_url FROM instances WHERE hash = ? OR domain LIKE ? LIMIT 1", [blocked_hash, domain.replace(char, "_")]
356 row = database.cursor.fetchone()
357 logger.debug("row[]='%s'", type(row))
360 logger.debug("blocked_hash='%s' not found, trying domain='%s' ...", blocked_hash, domain)
361 return deobfuscate(char, domain)
362 elif not domain.startswith("*."):
363 logger.debug("domain='%s' - BEFORE!", domain)
364 domain = tidyup.domain(domain)
365 logger.debug("domain='%s' - AFTER!", domain)
368 logger.warning("domain is empty after tidyup - EXIT!")
371 search = domain.replace(char, "_")
373 logger.debug("Looking up domain='%s',search='%s' ...", domain, search)
374 database.cursor.execute(
375 "SELECT domain, origin, nodeinfo_url FROM instances WHERE domain LIKE ? OR 'https://' || domain LIKE ? ORDER BY rowid LIMIT 1", [search, search]
378 row = database.cursor.fetchone()
379 logger.debug("row[]='%s'", type(row))
381 logger.debug("row[]='%s' - EXIT!", type(row))
384 def set_last_blocked(domain: str):
385 logger.debug("domain='%s' - CALLED!", domain)
386 domain_helper.raise_on(domain)
389 _set_data("last_blocked", domain, time.time())
390 logger.debug("EXIT!")
392 def set_last_instance_fetch(domain: str):
393 logger.debug("domain='%s' - CALLED!", domain)
394 domain_helper.raise_on(domain)
397 _set_data("last_instance_fetch", domain, time.time())
398 logger.debug("EXIT!")
400 def set_total_peers(domain: str, peers: list):
401 logger.debug("domain='%s',peers()=%d - CALLED!", domain, len(peers))
402 domain_helper.raise_on(domain)
404 if not isinstance(peers, list):
405 raise ValueError(f"Parameter peers[]='{type(peers)}' is not of type 'list'")
408 _set_data("total_peers", domain, len(peers))
409 logger.debug("EXIT!")
411 def set_total_blocks(domain: str, blocks: list):
412 logger.debug("domain='%s',blocks()=%d - CALLED!", domain, len(blocks))
413 domain_helper.raise_on(domain)
415 if not isinstance(blocks, list):
416 raise ValueError(f"Parameter blocks[]='{type(blocks)}' is not of type 'list'")
419 _set_data("total_blocks", domain, len(blocks))
420 logger.debug("EXIT!")
422 def set_obfuscated_blocks(domain: str, obfuscated: int):
423 logger.debug("domain='%s',obfuscated=%d - CALLED!", domain, obfuscated)
424 domain_helper.raise_on(domain)
426 if not isinstance(obfuscated, int):
427 raise ValueError(f"Parameter obfuscated[]='{type(obfuscated)}' is not of type 'int'")
429 raise ValueError(f"Parameter obfuscated={obfuscated} is not valid")
432 _set_data("obfuscated_blocks", domain, obfuscated)
433 logger.debug("EXIT!")
435 def set_nodeinfo_url(domain: str, url: str):
436 logger.debug("domain='%s',url='%s' - CALLED!", domain, url)
437 domain_helper.raise_on(domain)
439 if not isinstance(url, str) and url is not None:
440 raise ValueError(f"Parameter url[]='{type(url)}' is not of type 'str'")
442 raise ValueError("Parameter 'url' is empty")
445 _set_data("nodeinfo_url", domain, url)
446 logger.debug("EXIT!")
448 def set_detection_mode(domain: str, mode: str):
449 logger.debug("domain='%s',mode='%s' - CALLED!", domain, mode)
450 domain_helper.raise_on(domain)
452 if not isinstance(mode, str) and mode is not None:
453 raise ValueError(f"Parameter mode[]='{type(mode)}' is not of type 'str'")
455 raise ValueError("Parameter 'mode' is empty")
458 _set_data("detection_mode", domain, mode)
459 logger.debug("EXIT!")
461 def set_has_obfuscation(domain: str, status: bool):
462 logger.debug("domain='%s',status='%s' - CALLED!", domain, status)
463 domain_helper.raise_on(domain)
465 if not isinstance(status, bool):
466 raise ValueError(f"Parameter status[]='{type(status)}' is not of type 'bool'")
469 _set_data("has_obfuscation", domain, status)
470 logger.debug("EXIT!")
472 def set_software(domain: str, software: str):
473 logger.debug("domain='%s',software='%s' - CALLED!", domain, software)
474 domain_helper.raise_on(domain)
476 if not isinstance(software, str) and software is not None:
477 raise ValueError(f"Parameter software[]='{type(software)}' is not of type 'str'")
479 raise ValueError("Parameter 'software' is empty")
482 _set_data("software", domain, software)
483 logger.debug("EXIT!")
485 def valid(value: str, column: str) -> bool:
486 logger.debug("value='%s' - CALLED!", value)
487 if not isinstance(value, str):
488 raise ValueError(f"Parameter value[]='{type(value)}' is not of type 'str'")
490 raise ValueError("Parameter 'value' is empty")
491 elif not isinstance(column, str):
492 raise ValueError(f"Parameter column[]='{type(column)}' is not of type 'str'")
494 raise ValueError("Parameter 'column' is empty")
497 database.cursor.execute(
498 f"SELECT {column} FROM instances WHERE {column} = ? LIMIT 1", [value]
501 is_valid = database.cursor.fetchone() is not None
503 logger.debug("is_valid='%s' - EXIT!", is_valid)
506 def translate_idnas(rows: list, column: str):
507 logger.debug("rows[]='%s' - CALLED!", type(rows))
508 if not isinstance(rows, list):
509 raise ValueError("rows[]='{type(rows)}' is not of type 'list'")
511 raise ValueError("Parameter 'rows' is an empty list")
512 elif not isinstance(column, str):
513 raise ValueError(f"column='{type(column)}' is not of type 'str'")
515 raise ValueError("Parameter 'column' is empty")
516 elif column not in ["domain", "origin"]:
517 raise ValueError(f"column='{column}' is not supported")
519 logger.info("Checking/converting %d domain names ...", len(rows))
521 logger.debug("row[]='%s'", type(row))
523 translated = row[column].encode("idna").decode("utf-8")
524 logger.debug("translated='%s',row[%s]='%s'", translated, column, row[column])
526 if translated != row[column]:
527 logger.info("Translated row[%s]='%s' to '%s'", column, row[column], translated)
528 if is_registered(translated, True):
529 logger.warning("Deleting row[%s]='%s' as translated='%s' already exist", column, row[column], translated)
530 database.cursor.execute(f"DELETE FROM instances WHERE {column} = ? LIMIT 1", [row[column]])
532 database.cursor.execute(f"UPDATE instances SET {column} = ? WHERE {column} = ? LIMIT 1", [translated, row[column]])
534 logger.debug("Invoking commit() ...")
535 database.connection.commit()
537 logger.debug("EXIT!")