Check if actor has changed

Can sometimes happen if Person Update has not been received

cache.py

@@ -8,13 +8,14 @@ __status__ = "Production"
 
 import os
 import datetime
+from session import urlExists
 from utils import loadJson
 from utils import saveJson
 from utils import getFileCaseInsensitive
 
 
-def removePersonFromCache(baseDir: str, personUrl: str,
-                          personCache: {}) -> bool:
+def _removePersonFromCache(baseDir: str, personUrl: str,
+                           personCache: {}) -> bool:
     """Removes an actor from the cache
     """
     cacheFilename = baseDir + '/cache/actors/' + \
@@ -28,6 +29,24 @@ def removePersonFromCache(baseDir: str, personUrl: str,
         del personCache[personUrl]
 
 
+def checkForChangedActor(session, baseDir: str,
+                         httpPrefix: str, domainFull: str,
+                         personUrl: str, avatarUrl: str, personCache: {},
+                         timeoutSec: 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 avatarUrl:
+        return
+    if domainFull in avatarUrl:
+        return
+    if urlExists(session, avatarUrl, timeoutSec, httpPrefix, domainFull):
+        return
+    _removePersonFromCache(baseDir, personUrl, personCache)
+
+
 def storePersonInCache(baseDir: str, personUrl: str,
                        personJson: {}, personCache: {},
                        allowWriteToFile: bool) -> None:

daemon.py

@@ -20,7 +20,6 @@ import pyqrcode
 # for saving images
 from hashlib import sha256
 from hashlib import sha1
-from siteactive import siteIsActive
 from session import createSession
 from webfinger import webfingerMeta
 from webfinger import webfingerNodeInfo
@@ -229,7 +228,7 @@ from content import extractMediaInFormPOST
 from content import saveMediaInFormPOST
 from content import extractTextFieldsInPOST
 from media import removeMetaData
-from cache import removePersonFromCache
+from cache import checkForChangedActor
 from cache import storePersonInCache
 from cache import getPersonFromCache
 from httpsig import verifyPostHeaders
@@ -5491,14 +5490,11 @@ class PubServer(BaseHTTPRequestHandler):
                 if actorJson.get('alsoKnownAs'):
                     alsoKnownAs = actorJson['alsoKnownAs']
 
-            # check if the avatar image exists and if not then update
-            # the actor cache
-            if optionsProfileUrl:
-                if self.server.domainFull not in optionsProfileUrl:
-                    if not siteIsActive(optionsProfileUrl, 3):
-                        removePersonFromCache(self.server.baseDir,
-                                              optionsActor,
-                                              self.server.personCache)
+            checkForChangedActor(self.server.session,
+                                 self.server.baseDir, self.server.httpPrefix,
+                                 self.server.domainFull,
+                                 optionsActor, optionsProfileUrl,
+                                 self.server.personCache, 3)
 
             msg = htmlPersonOptions(self.server.defaultTimeline,
                                     self.server.cssCache,

session.py

@@ -53,6 +53,34 @@ def createSession(proxyType: str):
     return session
 
 
+def urlExists(session, url: str, timeoutSec=3,
+              httpPrefix='https', domain='testdomain') -> bool:
+    if not isinstance(url, str):
+        print('url: ' + str(url))
+        print('ERROR: urlExists failed, url should be a string')
+        return False
+    sessionParams = {}
+    sessionHeaders = {}
+    sessionHeaders['User-Agent'] = 'Epicyon/' + __version__
+    if domain:
+        sessionHeaders['User-Agent'] += \
+            '; +' + httpPrefix + '://' + domain + '/'
+    if not session:
+        print('WARN: urlExists failed, no session specified')
+        return True
+    try:
+        result = session.get(url, headers=sessionHeaders,
+                             params=sessionParams,
+                             timeout=timeoutSec)
+        if result:
+            if result.status_code == 200 or \
+               result.status_code == 304:
+                return True
+    except BaseException:
+        pass
+    return False
+
+
 def getJson(session, url: str, headers: {}, params: {},
             version='1.2.0', httpPrefix='https',
             domain='testdomain') -> {}:
@@ -72,6 +100,7 @@ def getJson(session, url: str, headers: {}, params: {},
             '; +' + httpPrefix + '://' + domain + '/'
     if not session:
         print('WARN: getJson failed, no session specified for getJson')
+        return None
     try:
         result = session.get(url, headers=sessionHeaders, params=sessionParams)
         return result.json()