#!/bin/bash

OUTPUT_FILE="tranco_V99PN_geolocation.csv"

# add ccTLDs
cat tranco_V99PN_ccTLDs.csv > $OUTPUT_FILE 

# add gTLDs with domain not found
cat tranco_V99PN_gTLDs_resolved.csv | grep "Error: (4," | sed "s/$/,unknown,unknown,unknown/" >> $OUTPUT_FILE
cat tranco_V99PN_gTLDs_resolved_2nd.csv | grep "Error: (4," | sed "s/$/,unknown,unknown,unknown/" >> $OUTPUT_FILE
cat tranco_V99PN_gTLDs_resolved_3rd.csv | grep "Error: (4," | sed "s/$/,unknown,unknown,unknown/" >> $OUTPUT_FILE

# add remaining gTLDs with other dns issues
cat tranco_V99PN_gTLDs_resolved_3rd.csv | grep "Error: (1," | sed "s/$/,unknown,unknown,unknown/" >> $OUTPUT_FILE 
cat tranco_V99PN_gTLDs_resolved_3rd.csv | grep "Error: (11," | sed "s/$/,unknown,unknown,unknown/" >> $OUTPUT_FILE
cat tranco_V99PN_gTLDs_resolved_3rd.csv | grep "Error: (12," | sed "s/$/,unknown,unknown,unknown/" >> $OUTPUT_FILE

# add gTLDs served by cdns
cat tranco_V99PN_gTLDs_resolved_final_cdn.csv | sed 's/-,-/unknown,unknown/' >> $OUTPUT_FILE

# Add gTLDs with geolocation
cat tranco_V99PN_gTLDs_resolved_final_nocdn_geolocation.csv | grep "," | grep "No," | sed 's/No,,/no valid ip,unknown,unknown/' >> $OUTPUT_FILE
cat tranco_V99PN_gTLDs_resolved_final_nocdn_geolocation.csv | grep "," | grep -v "No," | grep ",,," | sed 's/,,,/,private ip,unknown,unknown,/' >> $OUTPUT_FILE
cat tranco_V99PN_gTLDs_resolved_final_nocdn_geolocation.csv | grep "," | grep -v ",,," | grep -v ",No,," >> $OUTPUT_FILE

cat tranco_V99PN_geolocation.csv | sort -n > tranco_V99PN_geolocation_sorted.csv
