Cache blocked domains

blocking.py

@@ -9,6 +9,7 @@ __module_group__ = "ActivityPub"
 
 import os
 import json
+import time
 from datetime import datetime
 from utils import isAccountDir
 from utils import getCachedPostFilename
@@ -167,7 +168,29 @@ def getDomainBlocklist(baseDir: str) -> str:
     return blockedStr
 
 
-def isBlockedDomain(baseDir: str, domain: str) -> bool:
+def updateBlockedCache(baseDir: str,
+                       blockedCache: [],
+                       blockedCacheLastUpdated: int,
+                       blockedCacheUpdateSecs: int) -> int:
+    """Updates the cache of globally blocked domains held in memory
+    """
+    currTime = int(time.time())
+    secondsSinceLastUpdate = currTime - blockedCacheLastUpdated
+    if secondsSinceLastUpdate < blockedCacheUpdateSecs:
+        return blockedCacheLastUpdated
+    globalBlockingFilename = baseDir + '/accounts/blocking.txt'
+    if not os.path.isfile(globalBlockingFilename):
+        return blockedCacheLastUpdated
+    with open(globalBlockingFilename, 'r') as fpBlocked:
+        blockedLines = fpBlocked.readlines()
+        blockedCache.clear()
+        for line in blockedLines:
+            blockedCache.append(line)
+    return currTime
+
+
+def isBlockedDomain(baseDir: str, domain: str,
+                    blockedCache: [] = None) -> bool:
     """Is the given domain blocked?
     """
     if '.' not in domain:
@@ -186,16 +209,24 @@ def isBlockedDomain(baseDir: str, domain: str) -> bool:
 
     allowFilename = baseDir + '/accounts/allowedinstances.txt'
     if not os.path.isfile(allowFilename):
-        # instance block list
-        globalBlockingFilename = baseDir + '/accounts/blocking.txt'
-        if os.path.isfile(globalBlockingFilename):
-            with open(globalBlockingFilename, 'r') as fpBlocked:
-                blockedStr = fpBlocked.read()
+        if blockedCache:
+            for blockedStr in blockedCache:
                 if '*@' + domain in blockedStr:
                     return True
                 if shortDomain:
                     if '*@' + shortDomain in blockedStr:
                         return True
+        else:
+            # instance block list
+            globalBlockingFilename = baseDir + '/accounts/blocking.txt'
+            if os.path.isfile(globalBlockingFilename):
+                with open(globalBlockingFilename, 'r') as fpBlocked:
+                    blockedStr = fpBlocked.read()
+                    if '*@' + domain in blockedStr:
+                        return True
+                    if shortDomain:
+                        if '*@' + shortDomain in blockedStr:
+                            return True
     else:
         # instance allow list
         if not shortDomain:

daemon.py

@@ -114,6 +114,7 @@ from media import attachMedia
 from media import pathIsImage
 from media import pathIsVideo
 from media import pathIsAudio
+from blocking import updateBlockedCache
 from blocking import mutePost
 from blocking import unmutePost
 from blocking import setBrochMode
@@ -487,6 +488,12 @@ class PubServer(BaseHTTPRequestHandler):
         # is the User-Agent domain blocked
         blockedUA = False
         if not agentDomain.startswith(callingDomain):
+            self.server.blockedCacheLastUpdated = \
+                updateBlockedCache(self.server.baseDir,
+                                   self.server.blockedCache,
+                                   self.server.blockedCacheLastUpdated,
+                                   self.server.blockedCacheUpdateSecs)
+
             blockedUA = isBlockedDomain(self.server.baseDir, agentDomain)
             # if self.server.debug:
             if blockedUA:
@@ -1212,6 +1219,13 @@ class PubServer(BaseHTTPRequestHandler):
 
         messageDomain, messagePort = \
             getDomainFromActor(messageJson['actor'])
+
+        self.server.blockedCacheLastUpdated = \
+            updateBlockedCache(self.server.baseDir,
+                               self.server.blockedCache,
+                               self.server.blockedCacheLastUpdated,
+                               self.server.blockedCacheUpdateSecs)
+
         if isBlockedDomain(self.server.baseDir, messageDomain):
             print('POST from blocked domain ' + messageDomain)
             self._400()
@@ -15180,6 +15194,16 @@ def runDaemon(userAgentsBlocked: [],
     # contains threads used to send posts to followers
     httpd.followersThreads = []
 
+    # create a cache of blocked domains in memory.
+    # This limits the amount of slow disk reads which need to be done
+    httpd.blockedCache = []
+    httpd.blockedCacheLastUpdated = 0
+    httpd.blockedCacheUpdateSecs = 60
+    httpd.blockedCacheLastUpdated = \
+        updateBlockedCache(baseDir, httpd.blockedCache,
+                           httpd.blockedCacheLastUpdated,
+                           httpd.blockedCacheUpdateSecs)
+
     # cache to store css files
     httpd.cssCache = {}