fiex 4 and 10, sorting csv's now and making sure the origin of duplication is always local instance if compared

scripts/utils.py

@@ -36,7 +36,8 @@ def get_current_instance():
         return ""
 
 #Set the local dir on script run
-conf.LOCAL_DIR = conf.PARTICIPANT_DIR + get_current_instance() + '/'
+conf.LOCAL_INSTANCE = get_current_instance()
+conf.LOCAL_DIR = conf.PARTICIPANT_DIR + conf.LOCAL_INSTANCE + '/'
 
 ###################### Validations ######################
 
@@ -356,6 +357,37 @@ def merge_verification_df(receiving_df, merging_df):
     except Exception as err:
         return receiving_df
 
+def sort_instances(df, sort_by, preferred=None):
+    """
+    Sorts dataframe
+
+    Parameters
+    ----------
+        df : pd.DataFrame
+            The dataframe to sort
+        sort_by : str
+            The column to sort by
+        preferred(optional) : str
+            the preferred value is if i want the column to be sorted with a preferred value at the start of the dataframe
+
+    Returns
+    -------
+        pd.DataFrame
+            The sorted dataframe
+    """
+
+    try:
+        df = df.sort_values(by=sort_by)
+
+        if preferred:
+            df['priority'] = (df[sort_by] == preferred).astype(int)
+            df = df.sort_values(by=['priority', sort_by], ascending=[False, True]).drop(columns='priority')
+
+    except Exception as err:
+        print_colors('[-] Sorting failed',is_error=True)
+
+    return df
+
 def remove_duplications(df):
     """
     Remove url and name duplications from the dataframe
@@ -370,9 +402,10 @@ def remove_duplications(df):
         pd.DataFrame
             The dataframe after all duplications were removed
     """
+
     try:
-        df = df.drop_duplicates(subset='Name')
-        df = df.drop_duplicates(subset='URL')
+        df = df.drop_duplicates(subset='Name', keep='first')
+        df = df.drop_duplicates(subset='URL', keep='first')
 
     except Exception as err:
         print_colors('[-] Removing duplication failed',is_error=True)
@@ -704,10 +737,8 @@ def get_official_participants():
     """
 
     try:
-        current_instance = get_current_instance()
-
         with open(conf.OFFICIAL_PARTICIPANTS_FILE, 'r') as file:
-            return [line.strip() for line in file if current_instance not in line]
+            return [line.strip() for line in file if conf.LOCAL_INSTANCE not in line]
 
     except Exception as err:
         print_colors('[-] Couldn\'t read official webring participants file',is_error=True )