troll-patrol-artifact/scripts/belarus.sh

#!/bin/bash

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

cd belarus-2020-2021

if [ "$1" == "--fast" ]; then
    echo "Repacking some pre-processed data for the next step..."
    ./scripts/repack-preprocessed-data.sh
else
    echo "Downloading data from the Tor Project..."
    ./scripts/download-bridge-data.sh

    # Do this step in parallel
    for i in ${months[@]}; do
        if [ ! -e data/bridge-extra-infos-${i}_processed.tar.xz ]; then
            ./scripts/process-data-docker.sh "$i" &
        fi
    done
    wait
fi

docker run --name "tp-belarus" --rm -d -i tp-analysis:latest || exit 1
docker exec tp-belarus sh -c "mkdir -p /home/analysis/belarus-2020-2021/data/bridge_data"

# Copy archives to docker
for i in data/*_processed.tar.xz; do
    docker cp "${i}" tp-belarus:/home/analysis/belarus-2020-2021/data/bridge_data/
done

# Return to artifact root
cd ..

# Process the data
docker exec tp-belarus sh -c "cd /home/analysis/belarus-2020-2021 && ./scripts/get-email-bridges.sh && ./scripts/clean-bridge-data.sh" && \
docker exec tp-belarus sh -c "cd /home/analysis/belarus-2020-2021 && ./scripts/get-stats.sh" | tee appendix-a-results
docker cp tp-belarus:/home/analysis/belarus-2020-2021/appendix-a-results.tex . && \
docker cp tp-belarus:/home/analysis/belarus-2020-2021/appendix-a-results.pdf . && \
docker stop tp-belarus