Cache managed by redis now

.env.dist

@@ -5,11 +5,12 @@ MYSQL_USER=nyaa
 MYSQL_PASSWORD=nyaa
 MYSQL_DATABASE=nyaa
 MYSQL_SERVER=db
+REDIS_SERVER=redis
 ADMIN_USERNAME=admin
 ADMIN_PASSWORD=secret
 REQUESTS_TIMEOUT=5
 CACHE_TIMEOUT=3600
 MYSQL_ROOT_PASSWORD=root
-BLACKLIST_WORDS=Chris44,Vol.,[zza],.ssa,Ref:rain
+BLACKLIST_WORDS=Chris44,Vol.,[zza],.ssa,.ass,Ref:rain
 CLOUDPROXY_ENDPOINT=http://flaresolverr:8191/v1
 CAPTCHA_SOLVER=hcaptcha-solver

docker-compose.yml

@@ -26,6 +26,11 @@ services:
     volumes:
       - ./.db:/var/lib/mysql
 
+  redis:
+    image: redis
+    ports:
+      - "6379:6379"
+
   flaresolverr:
     image: flaresolverr/flaresolverr
     ports:

pynyaata/config.py

@@ -6,6 +6,7 @@ from flask.cli import load_dotenv
 from flask_apscheduler import APScheduler
 from flask_httpauth import HTTPBasicAuth
 from flask_sqlalchemy import SQLAlchemy
+from redis import Redis
 
 load_dotenv()
 
@@ -18,6 +19,7 @@ REQUESTS_TIMEOUT = int(environ.get('REQUESTS_TIMEOUT', 5))
 BLACKLIST_WORDS = environ.get('BLACKLIST_WORDS', '').split(',') if environ.get('BLACKLIST_WORDS', '') else []
 CLOUDPROXY_ENDPOINT = environ.get('CLOUDPROXY_ENDPOINT')
 MYSQL_ENABLED = False
+REDIS_ENABLED = False
 
 app = Flask(__name__)
 app.name = 'PyNyaaTa'
@@ -46,3 +48,8 @@ if db_host:
         'pool_recycle': 200
     }
     db = SQLAlchemy(app)
+
+cache_host = environ.get('REDIS_SERVER')
+if cache_host:
+    REDIS_ENABLED = True
+    cache = Redis(cache_host)

pynyaata/connectors/animeultime.py

@@ -58,7 +58,6 @@ class AnimeUltime(ConnectorCore):
                             'href': href,
                             'name': url.get_text(),
                             'type': tds[1].get_text(),
-                            'date': parse_date(None),
                             'class': self.color if link_exist_in_db(href) else ''
                         })
             elif len(player) > 0:
@@ -71,7 +70,6 @@ class AnimeUltime(ConnectorCore):
                     'href': '%s/file-0-1/%s' % (self.base_url, player[0]['data-serie']),
                     'name': name[0].get_text(),
                     'type': ani_type[0].get_text().replace(':', ''),
-                    'date': parse_date(None),
                     'class': self.color if link_exist_in_db(href) else ''
                 })
 

pynyaata/connectors/core.py

@@ -1,5 +1,4 @@
 from abc import ABC, abstractmethod
-from datetime import datetime
 from enum import Enum
 from functools import wraps
 from json import dumps, loads
@@ -8,7 +7,10 @@ from urllib.parse import urlencode
 import requests
 from requests import RequestException
 
-from ..config import CACHE_TIMEOUT, REQUESTS_TIMEOUT, CLOUDPROXY_ENDPOINT, logger
+from ..config import CACHE_TIMEOUT, REQUESTS_TIMEOUT, CLOUDPROXY_ENDPOINT, logger, REDIS_ENABLED
+
+if REDIS_ENABLED:
+    from ..config import cache
 
 cloudproxy_session = None
 
@@ -24,48 +26,30 @@ class ConnectorLang(Enum):
 
 
 class Cache:
-    CACHE_DATA = {}
-
     def cache_data(self, f):
         @wraps(f)
         def wrapper(*args, **kwds):
             connector = args[0]
-            timestamp = datetime.now().timestamp()
+            key = 'pynyaata.%s.%s.%s.%s' % (connector.__class__.__name__, f.__name__, connector.query, connector.page)
 
-            for connector_class in list(self.CACHE_DATA):
-                for connector_func in list(self.CACHE_DATA[connector_class]):
-                    for connector_query in list(self.CACHE_DATA[connector_class][connector_func]):
-                        for connector_page in list(self.CACHE_DATA[connector_class][connector_func][connector_query]):
-                            if self.CACHE_DATA[connector_class][connector_func][connector_query][connector_page][
-                                'timeout'
-                            ] < timestamp:
-                                del self.CACHE_DATA[connector_class][connector_func][connector_query][connector_page]
+            if REDIS_ENABLED:
+                json = cache.get(key)
 
-            if connector.__class__.__name__ not in self.CACHE_DATA:
-                self.CACHE_DATA[connector.__class__.__name__] = {}
-            if f.__name__ not in self.CACHE_DATA[connector.__class__.__name__]:
-                self.CACHE_DATA[connector.__class__.__name__][f.__name__] = {}
-            if connector.query not in self.CACHE_DATA[connector.__class__.__name__][f.__name__]:
-                self.CACHE_DATA[connector.__class__.__name__][f.__name__][connector.query] = {}
-            if connector.page not in self.CACHE_DATA[connector.__class__.__name__][f.__name__][connector.query]:
-                self.CACHE_DATA[connector.__class__.__name__][f.__name__][connector.query][connector.page] = {
-                    'timeout': 0.0
-                }
-
-            cached_data = self.CACHE_DATA[connector.__class__.__name__][f.__name__][connector.query][connector.page]
-            if cached_data['timeout'] > timestamp:
-                connector.data = cached_data['data']
-                connector.is_more = cached_data['is_more']
-                connector.on_error = False
-                return
+                if json:
+                    data = loads(json)
+                    connector.data = data['data']
+                    connector.is_more = data['is_more']
+                    connector.on_error = False
+                    return
 
             ret = f(*args, **kwds)
-            if not connector.on_error:
-                self.CACHE_DATA[connector.__class__.__name__][f.__name__][connector.query][connector.page] = {
+
+            if not connector.on_error and REDIS_ENABLED:
+                cache.set(key, dumps({
                     'data': connector.data,
-                    'timeout': timestamp + CACHE_TIMEOUT,
                     'is_more': connector.is_more
-                }
+                }), CACHE_TIMEOUT)
+
             return ret
 
         return wrapper

pynyaata/connectors/nyaa.py

@@ -65,7 +65,7 @@ class Nyaa(ConnectorCore):
                         'comment': str(urls[0]).replace('/view/', self.base_url + '/view/') if has_comment else '',
                         'link': tds[2].decode_contents().replace('/download/', self.base_url + '/download/'),
                         'size': tds[3].get_text(),
-                        'date': parse_date(tds[4].get_text(), '%Y-%m-%d %H:%M'),
+                        'date': tds[4].get_text(),
                         'seeds': check_seeds,
                         'leechs': tds[6].get_text(),
                         'downloads': check_downloads,

pynyaata/utils.py

@@ -21,9 +21,9 @@ def parse_date(str_to_parse, date_format=''):
     else:
         date = parse(str_to_parse, date_formats=[date_format])
         if date:
-            return date
+            return date.isoformat(' ', 'minutes')
         else:
-            return datetime.fromtimestamp(0)
+            return datetime.fromtimestamp(0).isoformat(' ', 'minutes')
 
 
 def boldify(str_to_replace, keyword):

requirements.txt

@@ -9,3 +9,4 @@ requests==2.25.1
 beautifulsoup4==4.9.3
 python-dotenv==0.18.0
 dateparser==1.0.0
+redis==3.5.3