import pandas as pd

'''
IMPORTANT: Remove leading spaces from columns in csv file or pandas won't read
it correctly.
'''

df_proto = pd.read_csv('csv/team13_protocol.csv', index_col=0).fillna(0)
df_all = pd.read_csv('csv/team13_monthly.csv', index_col=0).fillna(0)

# Calculate percentage of packets with 1 (ICMP), 6 (TCP) or 17 (UDP) protocols
packets_percentage_proto = ((df_proto['6 · # Packets'].sum() +
                             df_proto['17 · # Packets'].sum() +
                             df_proto['1 · # Packets'].sum()) /
                            df_all['#packets'].sum()) * 100

# Calculate ratio of packets from all other protocols
packets_percentage_others = 100 - packets_percentage_proto
print('Percentage of "others" from all protocols: ', round(packets_percentage_others, 2))

# Calculate percentage of unique source IPs with 1 (ICMP), 6 (TCP) or 17 (UDP) protocols
uIPs_percentage_proto = ((df_proto['6 · # Unique Source IPs'].sum() +
                          df_proto['17 · # Unique Source IPs'].sum() +
                          df_proto['1 · # Unique Source IPs'].sum()) /
                         df_all['#unique_IP_sources'].sum()) * 100

# Calculate ratio of unique source IPs from all other protocols
uIPs_percentage_others = 100 - uIPs_percentage_proto
print('Percentage of "others" from all protocols: ', round(uIPs_percentage_others, 1))

# Calculate percentage of unique destination IPs with 1 (ICMP), 6 (TCP) or 17 (UDP) protocols
uIPd_percentage_proto = ((df_proto['6 · # Unique Destination IPs'].sum() +
                          df_proto['17 · # Unique Destination IPs'].sum() +
                          df_proto['1 · # Unique Destination IPs'].sum()) /
                         df_all['#unique_IP_destinations'].sum()) * 100

# Calculate ratio of unique destination IPs from all other protocols
uIPd_percentage_others = 100 - uIPd_percentage_proto
print('Percentage of "others" from all protocols: ', round(uIPd_percentage_others, 1))