diff --git a/scripts/lantern.py b/scripts/lantern.py index 633ef3c..43e5c8c 100644 --- a/scripts/lantern.py +++ b/scripts/lantern.py @@ -623,7 +623,9 @@ Maintenance: rows2delete= [] # it is an empty list at first for i,j in csvdf.iterrows(): row=csvdf.loc[i,:].values.tolist() - #print_colors(f"{row}") + # check the number of columns in said row, + # print('rowcolnum:',len(row),' colnum:',len(csvdf.columns)) + # print_colors(f"{row}") @@ -669,6 +671,9 @@ Maintenance: filter_vdf= vdf[vdf.URL.str.contains(filterterm,na=False)] #print('2)',filter_vdf) #print('3)',uvdf[uvdf.URL.str.contains(filterterm,na=False)] ) + uvdf = pd.read_csv(unverifiedcsvfile, on_bad_lines='skip') + # TODO DELETE ALL DUPLICATES OF UVDF ! + uvdf = uvdf.drop_duplicates(subset=['URL'], keep="first", inplace=False) filter_uvdf= uvdf[uvdf.URL.str.contains(filterterm,na=False)] if len(filter_uvdf.index) == 0 and len(filter_vdf.index) == 0: newrow=row @@ -676,6 +681,7 @@ Maintenance: uvdf.index = uvdf.index + 1 # shifting index uvdf = uvdf.sort_index() # sorting by index uvdf.to_csv(unverifiedcsvfile, index=False) + print("[+] NEW ROW =",newrow) print_colors("[+] New row added to your own unverified.csv file!") else: