Tune performance

- Added threading to gunicorn instance
- Changed cache to cachetools's TTL cache since it uses LRU if
the cache is full
- Increased cache sizes 1.5x

.gitignore

@@ -100,3 +100,5 @@ $RECYCLE.BIN/
 
 # Windows shortcuts
 *.lnk
+
+*.gcloudignore

TwitchRSS/app.yaml

@@ -1,6 +1,6 @@
 runtime: python38
 
-entrypoint: gunicorn -b :$PORT twitchrss:app
+entrypoint: gunicorn -b :$PORT -k gthread --threads 2 twitchrss:app
 
 env_variables:
   TWITCH_CLIENT_ID: __INSERT_TWITCH_CLIENT_ID_HERE__

TwitchRSS/requirements.txt

@@ -1,5 +1,6 @@
 appdirs==1.4.3
 CacheControl==0.12.6
+cachetools==4.1.1
 certifi==2019.11.28
 chardet==3.0.4
 click==7.1.2
@@ -7,7 +8,6 @@ colorama==0.4.3
 contextlib2==0.6.0
 distlib==0.3.0
 distro==1.4.0
-expiringdict==1.2.1
 Flask==1.1.2
 gunicorn==20.0.4
 html5lib==1.0.1

TwitchRSS/twitchrss.py

@@ -22,14 +22,14 @@ import logging
 import re
 from os import environ
 from feedformatter import Feed
-from expiringdict import ExpiringDict
+from cachetools import cached, TTLCache, LRUCache
 from io import BytesIO
 import gzip
 
 
 VOD_URL_TEMPLATE = 'https://api.twitch.tv/kraken/channels/%s/videos?broadcast_type=archive,highlight,upload&limit=10'
 USERID_URL_TEMPLATE = 'https://api.twitch.tv/kraken/users?login=%s'
-VODCACHE_LIFETIME = 600
+VODCACHE_LIFETIME = 10 * 60
 USERIDCACHE_LIFETIME = 24 * 60 * 60
 CHANNEL_FILTER = re.compile("^[a-zA-Z0-9_]{2,25}$")
 TWITCH_CLIENT_ID = environ.get("TWITCH_CLIENT_ID")
@@ -41,9 +41,6 @@ if not TWITCH_CLIENT_ID:
 
 app = Flask(__name__)
 
-vodcache = ExpiringDict(max_len=200, max_age_seconds=VODCACHE_LIFETIME)
-useridcache = ExpiringDict(max_len=1000, max_age_seconds=USERIDCACHE_LIFETIME)
-
 
 @app.route('/vod/<string:channel>', methods=['GET', 'HEAD'])
 def vod(channel):
@@ -63,31 +60,29 @@ def vodonly(channel):
 
 def get_inner(channel, add_live=True):
     userid_json = fetch_userid(channel)
+    if not userid_json:
+        abort(404)
+
     (channel_display_name, channel_id) = extract_userid(json.loads(userid_json))
+
     channel_json = fetch_vods(channel_id)
+    if not channel_json:
+        abort(404)
+
     decoded_json = json.loads(channel_json)
     rss_data = construct_rss(channel, decoded_json, channel_display_name, add_live)
     headers = {'Content-Type': 'application/rss+xml'}
     return rss_data, headers
 
 
+@cached(cache=TTLCache(maxsize=2000, ttl=USERIDCACHE_LIFETIME))
 def fetch_userid(channel_name):
-    return fetch_or_cache_object(channel_name, useridcache, USERID_URL_TEMPLATE)
+    return fetch_json(channel_name, USERID_URL_TEMPLATE)
 
 
+@cached(cache=TTLCache(maxsize=400, ttl=VODCACHE_LIFETIME))
 def fetch_vods(channel_id):
-    return fetch_or_cache_object(channel_id, vodcache, VOD_URL_TEMPLATE)
-
-
-def fetch_or_cache_object(key, cachedict, url_template):
-    json_data = cachedict.get(key)
-    if not json_data:
-        json_data = fetch_json(key, url_template)
-        if not json_data:
-            abort(404)
-        else:
-            cachedict[key] = json_data
-    return json_data
+    return fetch_json(channel_id, VOD_URL_TEMPLATE)
 
 
 def fetch_json(id, url_template):
@@ -112,7 +107,7 @@ def fetch_json(id, url_template):
         except Exception as e:
             logging.warning("Fetch exception caught: %s" % e)
             retries += 1
-    return None
+    abort(503)
 
 
 def extract_userid(user_info):