belarus-2020-2021/scripts/get-bridge-data.sh

#!/bin/bash

parallel=false
if [ "$1" == "-p" ]; then
    parallel=true
fi

# Populate array of months we care about
months=()
# 2020
for i in $(seq 7 12); do
    months+=( 2020-$(printf %02d $i) )
done
# 2021
for i in $(seq 1 4); do
    months+=( 2021-$(printf %02d $i) )
done

# Download the archives if we don't have them already
for i in ${months[@]}; do
    if [ ! -f data/bridge-extra-infos-${i}.tar.xz ]; then
        curl -Lo data/bridge-extra-infos-${i}.tar.xz https://collector.torproject.org/archive/bridge-descriptors/extra-infos/bridge-extra-infos-${i}.tar.xz || exit 1
    fi
done

# Check that we have the right archives
sha256sum -c data/bridge-extra-infos.sha256 || exit 1

# Extract the data for each month
if [ "$parallel" == "true" ]; then
    # Do it in parallel
    for i in ${months[@]}; do
        ./scripts/extract-extra-infos-archive.sh bridge-extra-infos-${i}.tar.xz &
    done

    # Wait until we're done extracting everything
    wait
else
    for i in ${months[@]}; do
        ./scripts/extract-extra-infos-archive.sh bridge-extra-infos-${i}.tar.xz
    done
fi