cached-consensus-stats

#!/bin/bash -e

# Non-Fast&Stable relays are ignored
# BadExit flag disables Exit flag
# Valid, Running flags are ignored (all relays have them)
# Named, Unnamed, HSDir, V2Dir, Authority flags are ignored

insum() {
    local sum=0 x=
    while read x; do
        sum=$((sum + x))
    done
    echo ${sum}
}

cat cached-consensus \
    | sed -rn 's/^r ([^ ]+ ){5}([^ ]+).*/\2/p; /^s /s:(s? |$):/:gp; s/^w Bandwidth=(.*)/\1^/p' \
    | tr '\n^' ' \n' | sed 's/^ //' | sort -nrk3 \
    | grep '/Fast.*/Stable' \
    | sed -r 's:/(BadExit(/Exit)?|Fast|Stable|Named|Unnamed|Running|Valid|HSDir|V2Dir|Authority)::g' \
    > nodes

guards=`awk '/\/Guard/ {print $3}' nodes | insum`
exits=`awk '/\/Exit/ {print $3}' nodes | insum`
total=`awk '{print $3}' nodes | insum`

echo "Guards: ${guards}, Exits: ${exits}, Total: ${total}"

rm -f nodes-sniff
snsum=0
ord=0
cat nodes | while read ip flags bw; do
    if [ ${flags} = /Exit/Guard/ ]; then
        snsum=$((snsum + bw))
        gper=$(((snsum * 100 + 50) / guards))
        eper=$(((snsum * 100 + 50) / exits))

        # mark same-Class-C subnets, don't aggregate with non-Exit+Guard
        # (so lower bound, but doesn't seem to matter much)
        if grep -qs "^${ip%.*}\." nodes-sniff; then
            ip=" ${ip}"
        else
            ord=$((ord + 1))
        fi

        echo -e "${ip}\t\t\t${gper} ${eper}\t\t${ord} $((gper * eper))" >> nodes-sniff
    fi
done

cat nodes-sniff \
    | awk '{print $4,$5}' \
    | tac | sort -nuk1 \
    | sed 's/[0-9][0-9]$/.&%/' \
    > nodes-sniff-summary