Fix bug where dates were not indcluded when identifying duplicates

wiki_stats.py

@@ -86,15 +86,16 @@ def main(xml_name):
                 })
 
     df = pd.DataFrame(revisions)
-    df['date'] = pd.to_datetime(df['date'])
-    df = df.set_index('date').sort_index()
+    df.index = pd.to_datetime(df['date'])
 
     # Convert to local time, and round to nearest day
-    df.index = df.index.tz_convert('Australia/Sydney')
-    df.index = df.index.round('d')
+    df = df.tz_convert('Australia/Sydney')
+    df = df.sort_index()
+    df['date'] = df.index.date
 
     # Drop duplicate entries (ignore edits for same user/same page/same day)
     df = df.drop_duplicates()
+    df = df.drop(columns=['date'])
 
     # Only include non-academic users
     df = df[df['user'].isin(PROJECTS_TEAM)]