epicyon/cache.py

196 lines
7.0 KiB
Python
Raw Normal View History

2020-04-02 09:02:33 +00:00
__filename__ = "cache.py"
__author__ = "Bob Mottram"
__license__ = "AGPL3+"
2022-02-03 13:58:20 +00:00
__version__ = "1.3.0"
2020-04-02 09:02:33 +00:00
__maintainer__ = "Bob Mottram"
2021-09-10 16:14:50 +00:00
__email__ = "bob@libreserver.org"
2020-04-02 09:02:33 +00:00
__status__ = "Production"
2021-06-26 11:16:41 +00:00
__module_group__ = "Core"
2019-06-30 15:03:26 +00:00
2019-08-20 09:16:03 +00:00
import os
2019-06-30 15:18:40 +00:00
import datetime
2021-12-29 21:55:09 +00:00
from session import url_exists
from session import get_json
2021-12-26 15:13:34 +00:00
from utils import load_json
2021-12-26 14:47:21 +00:00
from utils import save_json
2021-12-29 21:55:09 +00:00
from utils import get_file_case_insensitive
2021-12-26 12:24:40 +00:00
from utils import get_user_paths
2020-04-02 09:02:33 +00:00
2020-05-04 19:16:11 +00:00
2021-12-30 18:38:36 +00:00
def _remove_person_from_cache(base_dir: str, person_url: str,
2021-12-29 21:55:09 +00:00
person_cache: {}) -> bool:
"""Removes an actor from the cache
"""
2021-12-30 18:38:36 +00:00
cache_filename = base_dir + '/cache/actors/' + \
person_url.replace('/', '#') + '.json'
if os.path.isfile(cache_filename):
try:
2021-12-30 18:38:36 +00:00
os.remove(cache_filename)
2021-11-25 18:42:38 +00:00
except OSError:
2021-12-30 18:38:36 +00:00
print('EX: unable to delete cached actor ' + str(cache_filename))
if person_cache.get(person_url):
del person_cache[person_url]
2021-12-29 21:55:09 +00:00
def check_for_changed_actor(session, base_dir: str,
http_prefix: str, domain_full: str,
person_url: str, avatar_url: str, person_cache: {},
2021-12-30 18:38:36 +00:00
timeout_sec: int):
"""Checks if the avatar url exists and if not then
the actor has probably changed without receiving an actor/Person Update.
So clear the actor from the cache and it will be refreshed when the next
post from them is sent
"""
if not session or not avatar_url:
return
if domain_full in avatar_url:
return
if url_exists(session, avatar_url, timeout_sec, http_prefix, domain_full):
return
2021-12-30 18:38:36 +00:00
_remove_person_from_cache(base_dir, person_url, person_cache)
2021-12-30 18:38:36 +00:00
def store_person_in_cache(base_dir: str, person_url: str,
person_json: {}, person_cache: {},
allow_write_to_file: bool) -> None:
2019-06-30 15:03:26 +00:00
"""Store an actor in the cache
"""
2021-12-30 18:38:36 +00:00
if 'statuses' in person_url or person_url.endswith('/actor'):
# This is not an actor or person account
return
2021-12-26 13:17:46 +00:00
curr_time = datetime.datetime.utcnow()
2021-12-30 18:38:36 +00:00
person_cache[person_url] = {
"actor": person_json,
2021-12-26 13:17:46 +00:00
"timestamp": curr_time.strftime("%Y-%m-%dT%H:%M:%SZ")
2019-07-06 17:00:22 +00:00
}
2021-12-25 16:17:53 +00:00
if not base_dir:
2019-08-20 09:16:03 +00:00
return
# store to file
2021-12-30 18:38:36 +00:00
if not allow_write_to_file:
2021-06-22 11:25:28 +00:00
return
2021-12-25 16:17:53 +00:00
if os.path.isdir(base_dir + '/cache/actors'):
2021-12-30 18:38:36 +00:00
cache_filename = base_dir + '/cache/actors/' + \
person_url.replace('/', '#') + '.json'
if not os.path.isfile(cache_filename):
save_json(person_json, cache_filename)
2020-04-02 09:02:33 +00:00
2019-06-30 15:03:26 +00:00
2021-12-30 18:38:36 +00:00
def get_person_from_cache(base_dir: str, person_url: str, person_cache: {},
allow_write_to_file: bool) -> {}:
2019-06-30 15:03:26 +00:00
"""Get an actor from the cache
"""
2019-08-20 09:37:09 +00:00
# if the actor is not in memory then try to load it from file
2021-12-30 18:38:36 +00:00
loaded_from_file = False
if not person_cache.get(person_url):
# does the person exist as a cached file?
2021-12-30 18:38:36 +00:00
cache_filename = base_dir + '/cache/actors/' + \
person_url.replace('/', '#') + '.json'
actor_filename = get_file_case_insensitive(cache_filename)
if actor_filename:
person_json = load_json(actor_filename)
if person_json:
store_person_in_cache(base_dir, person_url, person_json,
2021-12-29 21:55:09 +00:00
person_cache, False)
2021-12-30 18:38:36 +00:00
loaded_from_file = True
2020-03-22 21:16:02 +00:00
2021-12-30 18:38:36 +00:00
if person_cache.get(person_url):
if not loaded_from_file:
2019-08-20 09:50:27 +00:00
# update the timestamp for the last time the actor was retrieved
2021-12-26 13:17:46 +00:00
curr_time = datetime.datetime.utcnow()
2021-12-30 18:38:36 +00:00
curr_time_str = curr_time.strftime("%Y-%m-%dT%H:%M:%SZ")
person_cache[person_url]['timestamp'] = curr_time_str
return person_cache[person_url]['actor']
2019-08-20 09:37:09 +00:00
return None
2020-04-02 09:02:33 +00:00
2021-12-29 21:55:09 +00:00
def expire_person_cache(person_cache: {}):
2019-08-20 09:37:09 +00:00
"""Expires old entries from the cache in memory
"""
2021-12-26 13:17:46 +00:00
curr_time = datetime.datetime.utcnow()
2020-04-02 09:02:33 +00:00
removals = []
2021-12-30 18:38:36 +00:00
for person_url, cache_json in person_cache.items():
cache_time = datetime.datetime.strptime(cache_json['timestamp'],
"%Y-%m-%dT%H:%M:%SZ")
days_since_cached = (curr_time - cache_time).days
if days_since_cached > 2:
removals.append(person_url)
2020-04-02 09:02:33 +00:00
if len(removals) > 0:
2021-12-30 18:38:36 +00:00
for person_url in removals:
del person_cache[person_url]
2020-04-02 09:02:33 +00:00
print(str(len(removals)) + ' actors were expired from the cache')
2019-08-20 09:37:09 +00:00
2020-04-02 09:02:33 +00:00
2022-01-01 15:11:42 +00:00
def store_webfinger_in_cache(handle: str, webfing,
cached_webfingers: {}) -> None:
2019-08-20 09:37:09 +00:00
"""Store a webfinger endpoint in the cache
"""
2022-01-01 15:11:42 +00:00
cached_webfingers[handle] = webfing
2020-04-02 09:02:33 +00:00
2019-06-30 15:03:26 +00:00
2021-12-29 21:55:09 +00:00
def get_webfinger_from_cache(handle: str, cached_webfingers: {}) -> {}:
2019-06-30 15:03:26 +00:00
"""Get webfinger endpoint from the cache
"""
2021-12-25 22:28:18 +00:00
if cached_webfingers.get(handle):
return cached_webfingers[handle]
2019-06-30 15:03:26 +00:00
return None
2021-07-31 11:56:28 +00:00
2021-12-30 18:38:36 +00:00
def get_person_pub_key(base_dir: str, session, person_url: str,
2021-12-29 21:55:09 +00:00
person_cache: {}, debug: bool,
project_version: str, http_prefix: str,
domain: str, onion_domain: str,
i2p_domain: str,
2021-12-29 21:55:09 +00:00
signing_priv_key_pem: str) -> str:
2021-12-30 18:38:36 +00:00
if not person_url:
2021-07-31 11:56:28 +00:00
return None
2021-12-30 18:38:36 +00:00
person_url = person_url.replace('#main-key', '')
users_paths = get_user_paths()
for possible_users_path in users_paths:
if person_url.endswith(possible_users_path + 'inbox'):
2021-07-31 11:56:28 +00:00
if debug:
print('DEBUG: Obtaining public key for shared inbox')
2021-12-30 18:38:36 +00:00
person_url = \
person_url.replace(possible_users_path + 'inbox', '/inbox')
2021-07-31 11:56:28 +00:00
break
2021-12-30 18:38:36 +00:00
person_json = \
get_person_from_cache(base_dir, person_url, person_cache, True)
if not person_json:
2021-07-31 11:56:28 +00:00
if debug:
2021-12-30 18:38:36 +00:00
print('DEBUG: Obtaining public key for ' + person_url)
person_domain = domain
2021-12-25 20:43:43 +00:00
if onion_domain:
2021-12-30 18:38:36 +00:00
if '.onion/' in person_url:
person_domain = onion_domain
elif i2p_domain:
if '.i2p/' in person_url:
person_domain = i2p_domain
2021-12-30 18:38:36 +00:00
profile_str = 'https://www.w3.org/ns/activitystreams'
accept_str = \
'application/activity+json; profile="' + profile_str + '"'
as_header = {
'Accept': accept_str
2021-07-31 11:56:28 +00:00
}
2021-12-30 18:38:36 +00:00
person_json = \
2021-12-29 21:55:09 +00:00
get_json(signing_priv_key_pem,
2021-12-30 18:38:36 +00:00
session, person_url, as_header, None, debug,
project_version, http_prefix, person_domain)
if not person_json:
2021-07-31 11:56:28 +00:00
return None
2021-12-30 18:38:36 +00:00
pub_key = None
if person_json.get('publicKey'):
if person_json['publicKey'].get('publicKeyPem'):
pub_key = person_json['publicKey']['publicKeyPem']
2021-07-31 11:56:28 +00:00
else:
2021-12-30 18:38:36 +00:00
if person_json.get('publicKeyPem'):
pub_key = person_json['publicKeyPem']
2021-07-31 11:56:28 +00:00
2021-12-30 18:38:36 +00:00
if not pub_key:
2021-07-31 11:56:28 +00:00
if debug:
2021-12-30 18:38:36 +00:00
print('DEBUG: Public key not found for ' + person_url)
2021-07-31 11:56:28 +00:00
2021-12-30 18:38:36 +00:00
store_person_in_cache(base_dir, person_url, person_json,
person_cache, True)
return pub_key