Fix spam thread checking behavior

The program's false-positive check was interfering with the ability to catch many spam bot threads. This should remedy that.
ThioJoe · Oct 26, 2022 · c4b4f6a · c4b4f6a
1 parent 0d8a037
commit c4b4f6a
Show file tree

Hide file tree

Showing 2 changed files with 10 additions and 1 deletion.
diff --git a/Scripts/operations.py b/Scripts/operations.py
@@ -846,6 +846,7 @@ def check_against_filter(current, filtersDict, miscData, config, currentCommentD
 
       # Spam Lists
       spamListCombinedRegex = smartFilter['spamListCombinedRegex']
+      spamThreadsRegex = smartFilter['spamThreadsRegex']
 
       # if debugSingleComment == True: 
       #   if input("Sensitive True/False: ").lower() == 'true': sensitive = True
@@ -937,6 +938,10 @@ def multiVarDetect(text, username):
         processedText = commentText
       upLowTextSet = set(processedText)
 
+      # Run Spam Thread specific check first
+      if spamThreadsRegex.search(commentTextNormalized.lower()):
+        add_spam(current, config, miscData, currentCommentDict, videoID)
+
       # Run Checks
       if authorChannelID == parentAuthorChannelID:
         pass

diff --git a/Scripts/prepare_modes.py b/Scripts/prepare_modes.py
@@ -400,16 +400,19 @@ def prepare_filter_mode_smart(scanMode, config, miscData, sensitive=False):
   print("  Loading Filters  [===================           ]", end="\r")
 
   spamListExpressionsList = []
+  spamThreadsExpressionsList = []
   # Prepare spam domain regex
   for domain in spamDomainsList:
     spamListExpressionsList.append(confusable_regex(domain.upper().replace(".", "⚫"), include_character_padding=False).replace("(?:⚫)", "(?:[^a-zA-Z0-9 ]{1,2})"))
   for account in spamAccountsList:
     spamListExpressionsList.append(confusable_regex(account.upper(), include_character_padding=True).replace(m, a))
   for spamName in spamThreadsList:
     #spamListExpressionsList.append(confusable_regex(thread.upper(), include_character_padding=True).replace(m, a)) #With Confusables
-    spamListExpressionsList.append(re.escape(spamName.lower())) #Exact lowercase match
+    spamThreadsExpressionsList.append(re.escape(spamName.lower())) #Exact lowercase match
   print("  Loading Filters  [======================        ]", end="\r")
   spamListCombinedRegex = re.compile('|'.join(spamListExpressionsList))
+  print("  Loading Filters  [=========================     ]", end="\r")
+  spamThreadsRegex = re.compile('|'.join(spamThreadsExpressionsList))
 
   # Prepare Multi Language Detection
   turkish = 'ÇçŞşĞğİ'
@@ -445,6 +448,7 @@ def prepare_filter_mode_smart(scanMode, config, miscData, sensitive=False):
     'sensitiveRootDomainRegex': sensitiveRootDomainRegex,
     'unicodeCategoriesStrip': unicodeCategoriesStrip,
     'spamListCombinedRegex': spamListCombinedRegex,
+    'spamThreadsRegex': spamThreadsRegex,
     'threadFiltersDict': threadFiltersDict,
     'accompanyingLinkSpamDict': accompanyingLinkSpamDict,
     'comboDict': filter.comboDict